setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) ###### ######## Confirm genes => 18 ######### library(GSVA) library(GSEABase) library(msigdbr) library(clusterProfiler) library(org.Hs.eg.db) library(enrichplot) library(limma) library(tidyverse) Pathway <- read.gmt("../../msigdb/c2.all.v7.5.1.symbols.gmt") BCAA.SYMBOLs <- Pathway[str_detect(Pathway$term,"BRANCHED") | str_detect(Pathway$term,"VALINE"),] #### REACTOME 21 GENES #### BCAA.SYMBOLs[str_detect(BCAA.SYMBOLs$term,"REACTOME"),] %>% dim() #### KEGG 55 GENES #### BCAA.SYMBOLs[str_detect(BCAA.SYMBOLs$term,"KEGG"),] %>% dim() #### Total 56 GENES #### #### Venn #### dat<-c("REACTOME" = 21, "KEGG" = 55, "REACTOME&KEGG" = 18) library(Vennerable) x<-(BCAA.SYMBOLs[str_detect(BCAA.SYMBOLs$term,"REACTOME"),])$gene y<-(BCAA.SYMBOLs[str_detect(BCAA.SYMBOLs$term,"KEGG"),])$gene Figure.D<-Venn(list("REACTOME"=x,"KEGG"=y)) plot(Figure.D,doWeight=T) #### 18 GENES #### Symbols <- intersect(x,y) GeneType <- read.table("../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,Gene.name,NCBI.gene..formerly.Entrezgene..ID) %>% unique() data <- GeneType.Unique %>% filter(Gene.name %in% Symbols) %>% magrittr::set_colnames(c("ENSEMBL","SYMBOL","ENTREZ")) write.csv(data,"BCAA.ID.csv") ######## CNV ######### BCAA.ID <- read.csv("BCAA.ID.csv") library(readxl) CNV.Data <- read_xlsx("GSCA/CnvSummaryTable.xlsx") CNV.Data <- CNV.Data %>% mutate(cancertype=factor(cancertype,levels = rev(sort(unique(.$cancertype))))) Amp.Figure <- ggplot(CNV.Data,aes(symbol,cancertype,fill=a_total))+ geom_tile(color="black")+ geom_text(aes(label=round(a_total,1)),size=2)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title = "Amplication")+ scale_fill_gradientn("Percentage(%)",colors = brewer.pal(11,"RdBu")[6:1])+ theme(plot.title = element_text(hjust = 0.5)) ggsave(Amp.Figure,filename = "CNV.Amp.Figure.pdf",width = 5.5,height = 7) Del.Figure <- ggplot(CNV.Data,aes(symbol,cancertype,fill=d_total))+ geom_tile(color="black")+ geom_text(aes(label=round(d_total,1)),size=2)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title = "Deletion")+ scale_fill_gradientn("Percentage(%)",colors = brewer.pal(11,"RdBu")[6:11])+ theme(plot.title = element_text(hjust = 0.5)) ggsave(Del.Figure,filename = "CNV.Del.Figure.pdf",width = 5.5,height = 7) ######## CNV + Expression ######### library(tidyverse) library(readxl) library(RColorBrewer) C.Data <- read_xlsx("GSCA/CnvAndExpressionTable.xlsx") C.Data2 <- C.Data %>% mutate(symbol=factor(symbol,levels = rev(sort(unique(.$symbol))))) p1031.1 <- ggplot()+ geom_tile(data=C.Data2, aes(cancertype,symbol,fill=spm), color="white")+ geom_point(data=C.Data2 %>% filter(fdr > 0.05), aes(cancertype,symbol,size=-log10(fdr)), shape=4,color="black")+ geom_point(data=C.Data2 %>% filter(fdr <= 0.05), aes(cancertype,symbol,color=spm,size=-log10(fdr)), shape=1,color="black")+ scale_size_continuous("-log10(FDR)",range = c(0.5,4))+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradient2("Spearman'CC", high = "#E64B35FF", #brewer.pal(11,"RdBu")[1], low = "#053061", #brewer.pal(11,"RdBu")[11], mid = "white", breaks = c(-0.4,0,0.4,0.8)) data3 <- C.Data2 %>% mutate(Label=if_else(fdr>0.05,"P>0.05",if_else(spm>0,"Positive","Negative"))) %>% group_by(symbol,Label) %>% dplyr::summarise(Count=n()) %>% #mutate(symbol=factor(symbol,levels = Gene.Symble)) %>% ungroup() %>% mutate(Label=factor(Label,levels = c("P>0.05","Positive","Negative"))) p1031.top <- ggplot(data3,aes(Count,symbol,fill=Label))+ geom_col()+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_blank(), axis.text.x = element_text(size=9), axis.ticks.y = element_blank(), axis.title = element_text(size=10), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="Count",y="")+ scale_fill_manual("Spearman'CC",values = c("grey50",brewer.pal(11,"RdBu")[c(2,10)])) p1031.F <- p1031.1 %>% aplot::insert_right(p1031.top,width = 0.1) ggsave(p1031.F,filename = "CNV.Expression.Spearman.Figure.pdf",width = 10,height = 3) ######## SNV ######### BCAA.ID <- read.csv("BCAA.ID.csv") library(maftools) library(ComplexHeatmap) library(circlize) library(tidyverse) library(data.table) #### Mutation Freq #### SNP.Data <- read_xlsx("GSCA/SnvSummaryTable.xlsx") SNP.Data <- SNP.Data %>% mutate(Count=EffectiveMut+NonEffectiveMut) %>% mutate(cancertype=paste(cancertype,"(n=",sample_size,")",sep = "")) %>% dplyr::select(cancertype,symbol,Count) %>% spread(symbol,Count,fill=0) %>% gather(symbol,Count,-cancertype) %>% mutate(SampleSize=str_remove_all(cancertype,".*=") %>% str_remove_all("\\)")) %>% mutate(SampleSize=as.numeric(as.character(SampleSize))) %>% mutate(Percentage=round(Count/SampleSize*100,2)) %>% mutate(cancertype=factor(cancertype,levels = rev(sort(unique(.$cancertype))))) write.csv(SNP.Data,file = "SNP.Frequency.csv",row.names = F) SNP.Figure <- ggplot(SNP.Data,aes(symbol,cancertype,fill=Percentage))+ geom_tile(color="black")+ geom_text(aes(label=round(Count,1)),size=2)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title = "Mutation")+ scale_fill_gradientn("Mutation frequency(%)",colors = brewer.pal(11,"BrBG")[6:11])+ theme(plot.title = element_text(hjust = 0.5)) ggsave(SNP.Figure,filename = "SNP.Frequency.Figure.pdf",width = 6,height = 7) #### Oncoplot => UCEC #### SNP.Data <- read.csv("../AnimoAcidSensing/TCGA.PanCancer.Symbol.MUTECT2.SNP.MAF.csv") SNP.Clinical <- read.csv("../AnimoAcidSensing/TCGA.PanCancer.Symbol.MUTECT2.SNP.MAF.Clinical.csv") UCEC.Clinical <- SNP.Clinical %>% filter(Cancer=="UCEC") UCEC.SNP <- SNP.Data %>% filter(Tumor_Sample_Barcode %in% UCEC.Clinical$Tumor_Sample_Barcode) UCEC.SNP2 <- UCEC.SNP %>% mutate(Variant_Classification = if_else(Variant_Classification == "inframe_insertion","In_Frame_Ins", if_else(Variant_Classification == "inframe_deletion","In_Frame_Del", if_else(Variant_Classification == "frameshift_variant","Frame_Shift_Var", if_else(Variant_Classification == "missense_variant","Missense_Mutation", if_else(Variant_Classification == "synonymous_variant","Nonsense_Mutation", if_else(Variant_Classification == "3_prime_UTR_variant" | Variant_Classification == "5_prime_UTR_variant","Prime_UTR_Var", if_else(Variant_Classification == "intron_variant","Intron_Var", if_else(str_detect(Variant_Classification,";"),"Multi_Hit", if_else(str_detect(Variant_Classification,"splice"),"Splice_Site","Multi_Hit")))))))))) library(maftools) library(ComplexHeatmap) library(circlize) library(tidyverse) library(data.table) tcga = read.maf(maf = UCEC.SNP2,clinicalData=UCEC.Clinical, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) # library(RColorBrewer) vc_cols = brewer.pal(9,"Paired") names(vc_cols) <- c("In_Frame_Ins", "Missense_Mutation", "In_Frame_Del", "Frame_Shift_Var", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site", "Multi_Hit" ) SYMBOLs <- (BCAA.ID[BCAA.ID$SYMBOL %in% UCEC.SNP2$Hugo_Symbol,])$SYMBOL maftools::oncoplot(maf = tcga,genes = SYMBOLs,bgCol = "white", colors = vc_cols, additionalFeatureCol = "white", showTitle =F, annoBorderCol = "white", clinicalFeatures=c("OS")) #### Oncoplot => SKCM #### SKCM.Clinical <- SNP.Clinical %>% filter(Cancer=="SKCM") SKCM.SNP <- SNP.Data %>% filter(Tumor_Sample_Barcode %in% SKCM.Clinical$Tumor_Sample_Barcode) SKCM.SNP2 <- SKCM.SNP %>% mutate(Variant_Classification = if_else(Variant_Classification == "inframe_insertion","In_Frame_Ins", if_else(Variant_Classification == "inframe_deletion","In_Frame_Del", if_else(Variant_Classification == "frameshift_variant","Frame_Shift_Var", if_else(Variant_Classification == "missense_variant","Missense_Mutation", if_else(Variant_Classification == "synonymous_variant","Nonsense_Mutation", if_else(Variant_Classification == "3_prime_UTR_variant" | Variant_Classification == "5_prime_UTR_variant","Prime_UTR_Var", if_else(Variant_Classification == "intron_variant","Intron_Var", if_else(str_detect(Variant_Classification,";"),"Multi_Hit", if_else(str_detect(Variant_Classification,"splice"),"Splice_Site","Multi_Hit")))))))))) library(maftools) library(ComplexHeatmap) library(circlize) library(tidyverse) library(data.table) tcga = read.maf(maf = SKCM.SNP2,clinicalData=SKCM.Clinical, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) # library(RColorBrewer) vc_cols = brewer.pal(9,"Paired") names(vc_cols) <- c("In_Frame_Ins", "Missense_Mutation", "In_Frame_Del", "Frame_Shift_Var", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site", "Multi_Hit" ) SYMBOLs <- (BCAA.ID[BCAA.ID$SYMBOL %in% UCEC.SNP2$Hugo_Symbol,])$SYMBOL maftools::oncoplot(maf = tcga,genes = SYMBOLs,bgCol = "white", colors = vc_cols, additionalFeatureCol = "white", showTitle =F, annoBorderCol = "white", clinicalFeatures=c("OS")) ######## Diagnosis ####### #### Logistic model => ≥5 Normal #### library(pROC) library(ROCR) library(caret) library(tidyverse) BCAA.ID <- read.csv("BCAA.ID.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) #files <- list.files("../../mRNA.PanCancer.Exp/") files <- list.files("../Pancancer/",pattern = "[DEG]",full.names = T) files <- files[str_detect(files,"DEG.csv")] dir = "../../Clinical.XENA/" Dia.AUC <- data.frame() for (file in files) { project <- str_remove(file,".*/") %>% str_remove_all("_.*") %>% str_remove_all(".*-") phenotype <- read.csv(paste(dir,project,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples %in% c("Primary Tumor","Primary Blood Derived Cancer - Peripheral Blood","Solid Tissue Normal")) TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM <- log2(TPM+1) Inter.Samples <- intersect(phenotype$submitter_id.samples,colnames(TPM)) Pheno.mid <- phenotype %>% filter(submitter_id.samples %in% Inter.Samples) %>% mutate(Label = ifelse(sample_type.samples %in% c("Primary Tumor","Primary Blood Derived Cancer - Peripheral Blood"),1,0)) %>% dplyr::select(submitter_id.samples,Label) TPM.Target <- TPM[BCAA.ID$ENSEMBL,Pheno.mid$submitter_id.samples] %>% t() %>% data.frame(check.names = F) colnames(TPM.Target) <- BCAA.ID$SYMBOL TPM.Target$Label = Pheno.mid$Label if (sum(TPM.Target$Label == 0) >= 10) { Input.Formula <- as.formula(paste("Label ~ ",paste0(BCAA.ID$SYMBOL,collapse = "+"),sep = "")) TPM.Target$Label <- factor(TPM.Target$Label,levels = c(0,1),labels = c("Normal","Tumor")) for (i in 1:500) { print(project) print(i) index <- createDataPartition(TPM.Target$Label, p=0.7, list = F) train <- TPM.Target[index, ] test <- TPM.Target[-index, ] control = trainControl(method = "cv",number = 2,classProbs = TRUE) glm.model = train(Input.Formula, data= train, method = "glm", metric = "ROC", trControl = control) glm.probs = predict(glm.model,test,type = "prob") glm.ROC = roc(response = test[,c("Label")], predictor = glm.probs$Tumor, levels = levels(test[,c("Label")])) AUC <- glm.ROC$auc %>% str_remove(".* ") %>% as.numeric() %>% round(.,4) Dia.AUC <- data.frame(AUC = AUC, Repeat = i) %>% mutate(Cancer = project) %>% rbind.data.frame(Dia.AUC) } }else{ print(project) } } write.csv(Dia.AUC,file = "Diagnosis.BCAA.csv",row.names = T) setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) data <- read.csv("Diagnosis.BCAA.csv",row.names = 1) %>% group_by(Cancer) %>% mutate(AUC.Mean=mean(AUC), AUC.Median=median(AUC), AUC.SD=sd(AUC)) %>% dplyr::select(-Repeat,-AUC) %>% unique() read.csv("Diagnosis.BCAA.csv",row.names = 1) %>% write.csv(file="TableS1.csv") write.csv(data,file="TableS1.Sup.csv") #### Figure #### library(tidyverse) Dia.AUC <- read.csv("Diagnosis.BCAA.csv",row.names = 1) #Dia.AUC <- Dia.AUC %>% # mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) library(RColorBrewer) Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[1:3], brewer.pal(8,"Dark2")[6], brewer.pal(12,"Set3")[4:12], brewer.pal(8,"Dark2")[1:4]) p1104 <- ggplot(Dia.AUC,aes(x=AUC,y=reorder(Cancer,AUC),fill=Cancer))+ #geom_violin(cex=1.2)+ geom_boxplot()+ #geom_jitter() ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="AUROC",y="")+ scale_fill_manual(values = Cancer.Col) ggsave(p1104,filename = "Diagnosis.BCAA.pdf",width = 2,height = 4) ######## PCA + UMAP ######### library(tidyverse) BCAA.ID <- read.csv("BCAA.ID.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) #files <- list.files("../../mRNA.PanCancer.Exp/") files <- list.files("../Pancancer/",pattern = "[DEG]",full.names = T) files <- files[str_detect(files,"DEG.csv")] dir = "../../Clinical.XENA/" BCAA.Data <- data.frame() for (file in files) { project <- str_remove(file,".*/") %>% str_remove_all("_.*") %>% str_remove_all(".*-") phenotype <- read.csv(paste(dir,project,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples %in% c("Primary Tumor","Primary Blood Derived Cancer - Peripheral Blood","Solid Tissue Normal")) TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM <- log2(TPM+1) Inter.Samples <- intersect(phenotype$submitter_id.samples,colnames(TPM)) Pheno.mid <- phenotype %>% filter(submitter_id.samples %in% Inter.Samples) %>% mutate(Label = ifelse(sample_type.samples %in% c("Primary Tumor","Primary Blood Derived Cancer - Peripheral Blood"),1,0)) %>% dplyr::select(submitter_id.samples,Label) TPM.Target <- TPM[BCAA.ID$ENSEMBL,Pheno.mid$submitter_id.samples] %>% t() %>% data.frame(check.names = F) colnames(TPM.Target) <- BCAA.ID$SYMBOL TPM.Target$Label = Pheno.mid$Label BCAA.Data <- TPM.Target %>% rownames_to_column("SampleID") %>% mutate(Cancer=project) %>% rbind.data.frame(BCAA.Data) } BCAA.Data2 <- BCAA.Data %>% mutate(Type=if_else(Label==1,"Tumor","Normal")) write.csv(BCAA.Data2,file = "BCAA.PCA.csv",row.names = F) #### PCA #### BCAA.Data2 <- read.csv(file = "BCAA.PCA.csv") PCA.Data <- BCAA.Data2 %>% dplyr::select(-Label,-Cancer,-Type) %>% remove_rownames() %>% column_to_rownames("SampleID") PCA.Condition <- BCAA.Data2 %>% dplyr::select(SampleID,Label,Cancer,Type) library(ggbiplot) #data(wine) BCAA.PCA <- prcomp(PCA.Data, scale. = TRUE) PCA.Data <- BCAA.PCA$x[,1:5] %>% data.frame(check.names = F) %>% cbind.data.frame(PCA.Condition) write.csv(PCA.Data,file = "BCAA.PCA.Figure.Data.csv") Immune.Col <- c("#293462","#a64942","#fe5f55","#fff1c1", "#5bd1d7","#348498","#004d61","#ff502f", "#e41749","#f5587b","#ff8a5c","#fff591", "#001871","#ff585d","#ffb549","#41b6e6", "#515bd4","#8134af","#dd2a7b","#feda77", "#96ceb4","#ffeead","#d9534f","#ffad60", "#05445c","#f2317f","#5c4f74","#040000", "#de4307","#f29c2b","#f6d04d","#8bc24c", "#fef4a9","#3b9a9c","#4bc2c5","#78fee0", "#fad3cf","#a696c8","#2470a0","#c50d66", "#f07810","#eec60a","#dd7777","#1687a7", "#014955","#dd0a35","#ED5485","#FFE869", "#bc8420","#F0B775","#D25565","#2E94B9") ##Immune.Col[c(1:4,6:12,15,17,18,20,30:38)] p0105 <- ggplot(PCA.Data,aes(x=PC1,y=PC2,color=Cancer,shape=Type))+ geom_point(alpha=0.9)+ #geom_violin(cex=1.2)+ #geom_boxplot()+ #geom_jitter() ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="PC1(40.6%)",y="PC2(9.5%)")+ scale_color_manual(values = Immune.Col[c(1:4,6:12,15,17,18,20,30:38)]) ggsave(p0105,filename = "BCAA.PCA.Legend.pdf",width = 6,height = 4) ggsave(p0105,filename = "BCAA.PCA.pdf",width = 4,height = 2) #### UMAP #### BCAA.UMAP <- uwot::umap(PCA.Data, n_neighbors = 15, min_dist = 0.001, target_weight = 0.5) UMAP.Data <- BCAA.UMAP %>% data.frame() %>% cbind.data.frame(PCA.Condition) p0106 <- ggplot(UMAP.Data,aes(x=X1,y=X2,color=Cancer,shape=Type))+ geom_point(alpha=0.9)+ #geom_violin(cex=1.2)+ #geom_boxplot()+ #geom_jitter() ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="UMAP_1",y="UMAP_2")+ scale_color_manual(values = Immune.Col[c(1:4,6:12,15,17,18,20,30:38)]) ggsave(p0106,filename = "BCAA.UMAP.Legend.pdf",width = 6,height = 4) ggsave(p0106,filename = "BCAA.UMAP.pdf",width = 4,height = 2) ######## DEGs => TCGA => All Samples ########### BCAA.ID <- read.csv("BCAA.ID.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) #files <- list.files("../../mRNA.PanCancer.Exp/") files <- list.files("../Pancancer/",pattern = "[DEG]",full.names = T) files <- files[str_detect(files,"DEG.csv")] BCAA.DEG <- data.frame() for (file in files) { project = file %>% str_remove_all(".*\\/") %>% str_remove_all("_.*") %>% str_remove_all("TCGA-") data <- read.csv(file.path("../Pancancer/",file)) BCAA.DEG <- data %>% filter(X %in% BCAA.ID$ENSEMBL) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAA.DEG) } BCAA.DEG <- merge(BCAA.DEG,BCAA.ID,by.x="X",by.y = "ENSEMBL") write.csv(BCAA.DEG,file="BCAA.DEGs.csv") #### Figure #### BCAA.DEG <- read.csv("BCAA.DEGs.csv",row.names = 1) BCAA.DEG2 <- BCAA.DEG %>% mutate(Threshold=if_else(padj>0.5,"No", if_else(log2FoldChange>=1,"Up", if_else(log2FoldChange<=-1,"Down","No")))) %>% mutate(Label=cut(log2FoldChange,breaks = c(-Inf,-2,-1,0,1,2,Inf), labels = c("≤-2","-2~-1","-1~0","0~1","1~2","≥2"))) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) #%>% #mutate(SYMBOL=factor(SYMBOL,levels = )) BCAA.DEG2$Sig <- if_else(BCAA.DEG2$padj > 0.05,"P>0.05",as.character(BCAA.DEG2$Label)) BCAA.DEG2$Sig <- factor(BCAA.DEG2$Sig,levels = c("≤-2","-2~-1","-1~0","0~1","1~2","≥2","P>0.05")) library(RColorBrewer) p1030.2 <- ggplot(BCAA.DEG2,aes(SYMBOL,Cancer,fill=Sig))+ geom_tile(color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_manual("log2FoldChange",values = brewer.pal(11,"RdBu")[c(11,10,8,4,2,1,6)]) ggsave(p1030.2,filename = "BCAA.DEGs.Heatmap.pdf",height = 4,width = 6) ######## DEGs => TCGA => Paired Samples ####### library(readxl) Data <- read_xlsx("GSCA/DifferentialExpressionTable.xlsx") Data2 <- Data %>% mutate(log2FoldChange=log2(fc)) %>% mutate(Threshold=if_else(fdr>0.5,"No", if_else(log2FoldChange>=1,"Up", if_else(log2FoldChange<=-1,"Down","No")))) %>% mutate(Label=cut(log2FoldChange,breaks = c(-Inf,-2,-1,0,1,2,Inf), labels = c("≤-2","-2~-1","-1~0","0~1","1~2","≥2"))) %>% mutate(cancertype=factor(cancertype,levels = rev(sort(unique(.$cancertype))))) Data2$Sig <- if_else(Data2$fdr > 0.05,"P>0.05",as.character(Data2$Label)) Data2$Sig <- factor(Data2$Sig,levels = c("≤-2","-2~-1","-1~0","0~1","1~2","≥2","P>0.05")) library(RColorBrewer) p1030.3 <- ggplot(Data2,aes(symbol,cancertype,fill=Sig))+ geom_tile(color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_manual("log2FoldChange",values = brewer.pal(11,"RdBu")[c(11,10,8,4,2,1,6)]) ggsave(p1030.3,filename = "BCAA.DEGs.Heatmap.PairedSamples.pdf",height = 4,width = 6) ######## Survival => OS ########## library(tidyverse) library(readxl) library(survival) library(survminer) dir.create("Survival.KM") BCAA.ID <- read.csv("BCAA.ID.csv") dirs <- list.dirs("../../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] BCAA.KM.Pancancer <- data.frame() for (dir in dirs) { print(dir) project <- str_remove(dir,".*/") #dir.create(paste("AASS.KM/",project,sep = "")) ##submitter_id.samples sample_type.samples phenotype <- read.csv(paste(dir,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) #### Calculate AASS ssGSEA library(GSVA) BCAA=list() BCAA[["BCAA"]]<-BCAA.ID$ENSEMBL res.ssgsea <- gsva(as.matrix(TPM.log), gset.idx.list=BCAA, method = "ssgsea", kcdf = "Gaussian") #### survival datasets survival.data <- read.csv(paste(dir,"/TCGA-",project,".survival.tsv",sep = ""),sep = "\t",header = T,check.names = F) Inter.Samples <- intersect(intersect(phenotype$submitter_id.samples,colnames(TPM)),survival.data$sample) #### symbol expression TPM.ENSEMBL.Target <- TPM.log[BCAA.ID$ENSEMBL,Inter.Samples] %>% t() %>% data.frame(check.names = F) %>% mutate(SampleID = rownames(.)) %>% merge(.,survival.data,by.x="SampleID",by.y="sample") #### BCAA survival TPM.log.Target <- res.ssgsea %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("SampleID") %>% merge(TPM.ENSEMBL.Target,by="SampleID") %>% mutate(Cancer=project) #write.csv(TPM.log.Target,file = paste("AASS.KM/TCGA-",project,".AASS.SurvivalData.csv",sep = ""),row.names = F) res.cut <- surv_cutpoint(TPM.log.Target, time = paste("OS",".time",sep = ""), event = "OS", variables = "BCAA") #res.cat <- surv_categorize(res.cut) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(Surv(OS.time, OS) ~ BCAA, data = dat) index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","BCAA",sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ","BCAA",sep = "")), data = TPM.log.Target) coxSummary = summary(cox) BCAA.KM.Pancancer=rbind(BCAA.KM.Pancancer, data.frame(ENSEMBL="BCAA", KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project)) #### GENE SURVIVAL #OS.Tab=data.frame() for(i in BCAA.ID$ENSEMBL){ symbol <- BCAA.ID[BCAA.ID$ENSEMBL == i,]$SYMBOL %>% as.character() cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ",i,sep = "")), data = TPM.log.Target) coxSummary = summary(cox) #group=ifelse(TPM.log.Target[[i]] > median(TPM.log.Target[[i]]),"high","low") #if(length(table(group))==1) return(NULL) # res.cut <- surv_cutpoint(TPM.log.Target, time = paste("OS",".time",sep = ""), event = "OS", variables = i) #res.cat <- surv_categorize(res.cut) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(as.formula(paste("Surv(",index,".time,",index,") ~ ",i,sep = "")), data = dat) index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ",i,sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) #cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ",i,sep = "")), data = TPM.log.Target) #coxSummary = summary(cox) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) library(ggthemes) p<-ggsurvplot(fit,pval =TRUE, data = dat, surv.median.line = "hv", legend.title = symbol, conf.int.style = "step", xlab = "Time in days", #break.time.by = 500, risk.table = "abs_pct", #risk.table.y.text.col = T, #risk.table.y.text = FALSE, legend.labs = c("High", "Low"), #pval = TRUE, conf.int = TRUE, palette = "Set1", #ggtheme = theme.sur, risk.table.y.text.col = T, risk.table.y.text = F, ggtheme = theme.sur) print(p) #dev.off() graph2pdf(file=paste("Survival.KM/TCGA-",project,".OS.BestCutoff.KM.",symbol,".pdf",sep = ''),height = 5,width = 4) BCAA.KM.Pancancer=rbind(BCAA.KM.Pancancer, data.frame(ENSEMBL=symbol, KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project)) } } write.csv(BCAA.KM.Pancancer,file = "Survival.OS.BCAA.csv",row.names = F) if (F) { result.cox <- coxph(Surv(time, status) ~ sex, data = lung) summary(result.cox) ## Method1 data <- lung covariates <- c("age", "sex", "ph.karno", "ph.ecog", "wt.loss") results <- data.frame(covariate = character(0), beta = numeric(0), HR.confint.upper=numeric(0), HR.confint.upper=numeric(0), wald.test = numeric(0), p.value = numeric(0)) for (covariate in covariates) { formula <- as.formula(paste("Surv(time, status) ~", covariate)) model <- coxph(formula, data = data) summary <- summary(model) p.value <- signif(summary$wald["pvalue"], digits = 2) wald.test <- signif(summary$wald["test"], digits = 2) beta <- signif(summary$coef[1], digits = 2) HR.confint.lower <- signif(summary$conf.int[,"lower .95"], 2) HR.confint.upper <- signif(summary$conf.int[,"upper .95"], 2) results <- rbind(results, data.frame(covariate = covariate, beta = beta, HR.confint.lower=HR.confint.lower, HR.confint.upper=HR.confint.upper, wald.test= wald.test, p.value = p.value)) } #results ### Method2 data <- lung covariates <- c("age", "sex", "ph.karno", "ph.ecog", "wt.loss") univ_models <- map(covariates, function(covariate) { formula <- as.formula(paste("Surv(time, status) ~", covariate)) model <- coxph(formula, data = data) summary <- summary(model) p.value <- signif(summary$wald["pvalue"], digits = 2) wald.test <- signif(summary$wald["test"], digits = 2) beta <- signif(summary$coef[1], digits = 2) HR.confint.lower <- signif(summary$conf.int[,"lower .95"], 2) HR.confint.upper <- signif(summary$conf.int[,"upper .95"], 2) HR <- paste0(signif(exp(summary$coef[2]), digits = 2), " (", HR.confint.lower, "-", HR.confint.upper, ")") return(list(covariate = covariate, beta = beta, HR = HR, wald.test = wald.test, p.value = p.value)) }) results <- reduce(univ_models, rbind) results <- do.call(rbind, univ_models) } #### Figure #### library(tidyverse) library(RColorBrewer) BCAA.KM.Pancancer <- read.csv("Survival.OS.BCAA.csv") BCAA.KM.Pancancer2 <- BCAA.KM.Pancancer %>% mutate(Label = if_else(KM.pvalue > 0.05,"p>0.05", if_else(HR < 1,"Protective","Risky"))) %>% #mutate(ENSEMBL=factor(ENSEMBL,levels = rev(c("AASS",Gene.Symble)))) %>% mutate(Label=factor(Label,levels = c("Protective","Risky","p>0.05"))) %>% filter(ENSEMBL!="BCAA") %>% mutate(ENSEMBL=factor(ENSEMBL,levels = rev(sort(unique(.$ENSEMBL))))) p1 <- ggplot()+ geom_tile(data = BCAA.KM.Pancancer2, aes(y=ENSEMBL,x=Cancer,fill=Label), color="white")+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue>0.05), aes(y=ENSEMBL,x=Cancer,size=-log10(KM.pvalue)), shape=4,color="black")+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue<= 0.05), aes(y=ENSEMBL,x=Cancer,size=-log10(KM.pvalue)), shape=1,color="black")+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(KM p)",fill="Type")+ scale_fill_manual(values = c("#4DBBD5FF","#E5086A","grey90")) ggsave(p1,filename = "Survival.BCAA.OS.KM.pdf",height = 4,width = 8) BCAA.KM.Pancancer3 <- BCAA.KM.Pancancer %>% mutate(Label = if_else(HR.pvalue > 0.05,"p>0.05", if_else(HR < 1,"Protective","Risky"))) %>% #mutate(ENSEMBL=factor(ENSEMBL,levels = rev(c("AASS",Gene.Symble)))) %>% mutate(Label=factor(Label,levels = c("Protective","Risky","p>0.05"))) %>% filter(ENSEMBL!="BCAA") %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) p2 <- ggplot()+ geom_point(data = BCAA.KM.Pancancer3, aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(HR.pvalue)), shape=21,color="white",stroke=1)+ geom_point(data = BCAA.KM.Pancancer3 %>% filter(HR.pvalue<= 0.05), aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(HR.pvalue)), shape=21,color="black",stroke=1.2)+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(COX p)",fill="Type")+ scale_fill_manual(values = c("#469990","Orange","grey")) ggsave(p2,filename = "Survival.BCAA.OS.COX.pdf",height = 5,width = 5) ######## Survival => DSS PFI ######### library(tidyverse) library(readxl) library(survival) library(survminer) #dir.create("Survival.KM") BCAA.ID <- read.csv("BCAA.ID.csv") dirs <- list.dirs("../../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] BCAA.KM.Pancancer <- data.frame() for (dir in dirs[16:33]) { print(dir) project <- str_remove(dir,".*/") #dir.create(paste("AASS.KM/",project,sep = "")) ##submitter_id.samples sample_type.samples phenotype <- read.csv(paste(dir,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") %>% mutate(Sample.ID=substr(submitter_id.samples,1,15)) TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) #### Calculate AASS ssGSEA library(GSVA) BCAA=list() BCAA[["BCAA"]]<-BCAA.ID$ENSEMBL res.ssgsea <- gsva(as.matrix(TPM.log), gset.idx.list=BCAA, method = "ssgsea", kcdf = "Gaussian") #### survival datasets survival.data <- read.csv(paste(dir,"/../XENA.TCGA.",project,"_survival.txt",sep = ""), sep = "\t",header = T,check.names = F,row.names = 1) %>% rownames_to_column("Sample.ID") #Inter.Samples <- intersect(intersect(phenotype$submitter_id.samples,colnames(TPM)), # survival.data$sample) #### symbol expression TPM.ENSEMBL.Target <- TPM.log[BCAA.ID$ENSEMBL,] %>% t() %>% data.frame(check.names = F) %>% mutate(SampleID = rownames(.)) %>% mutate(Sample.ID=substr(SampleID,1,15)) %>% merge(.,survival.data,by.x="Sample.ID") %>% filter(Sample.ID %in% phenotype$Sample.ID) #### BCAA survival TPM.log.Target <- res.ssgsea %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("SampleID") %>% merge(TPM.ENSEMBL.Target,by="SampleID") %>% mutate(Cancer=project) #write.csv(TPM.log.Target,file = paste("AASS.KM/TCGA-",project,".AASS.SurvivalData.csv",sep = ""),row.names = F) #index="PFI" for (index in c("DSS","PFI")) { middata <- TPM.log.Target %>% dplyr::select(c("BCAA",BCAA.ID$ENSEMBL,"DSS","DSS.time","PFI","PFI.time")) TPM.log.Target2 <- middata %>% dplyr::select(c("BCAA",index,paste(index,".time",sep = ""))) %>% na.omit() if (nrow(TPM.log.Target2) >= 10) { res.cut <- surv_cutpoint(TPM.log.Target, time = paste(index,".time",sep = ""), event = index, variables = "BCAA") #res.cat <- surv_categorize(res.cut) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) #fit <- survfit(Surv(OS.time, OS) ~ BCAA,data = dat) #index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","BCAA",sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) cox <- coxph(as.formula(paste("Surv(",index,".time,",index,") ~ ","BCAA",sep = "")), data = TPM.log.Target) coxSummary = summary(cox) BCAA.KM.Pancancer=rbind(BCAA.KM.Pancancer, data.frame(ENSEMBL="BCAA", KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project, Type=index)) #### GENE SURVIVAL #OS.Tab=data.frame() for(i in BCAA.ID$ENSEMBL){ symbol <- BCAA.ID[BCAA.ID$ENSEMBL == i,]$SYMBOL %>% as.character() cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ",i,sep = "")), data = TPM.log.Target) coxSummary = summary(cox) #group=ifelse(TPM.log.Target[[i]] > median(TPM.log.Target[[i]]),"high","low") #if(length(table(group))==1) return(NULL) # res.cut <- surv_cutpoint(TPM.log.Target, time = paste(index,".time",sep = ""), event = index, variables = i) #res.cat <- surv_categorize(res.cut) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(as.formula(paste("Surv(",index,".time,",index,") ~ ",i,sep = "")), data = dat) #index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ",i,sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) #cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ",i,sep = "")), data = TPM.log.Target) #coxSummary = summary(cox) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) library(ggthemes) library(export) p<-ggsurvplot(fit,pval =TRUE, data = dat, surv.median.line = "hv", legend.title = symbol, conf.int.style = "step", xlab = "Time in days", #break.time.by = 500, risk.table = "abs_pct", #risk.table.y.text.col = T, #risk.table.y.text = FALSE, legend.labs = c("High", "Low"), #pval = TRUE, conf.int = TRUE, palette = "Set1", #ggtheme = theme.sur, risk.table.y.text.col = T, risk.table.y.text = F, ggtheme = theme.sur) print(p) #dev.off() graph2pdf(file=paste("Survival.KM/TCGA-",project,".",index,".BestCutoff.KM.",symbol,".pdf",sep = ''),height = 5,width = 4) BCAA.KM.Pancancer=rbind(BCAA.KM.Pancancer, data.frame(ENSEMBL=symbol, KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project, Type=index)) } } #write.csv(BCAA.KM.Pancancer,file = paste("Survival.",index,".BCAA.csv",sep = ""),row.names = F) } } BCAA.KM.Pancancer <- BCAA.KM.Pancancer %>% unique() write.csv(BCAA.KM.Pancancer,file = "Survival.DSS.PFI.BCAA.csv") #### Figure => DSS #### library(tidyverse) library(RColorBrewer) BCAA.KM.Pancancer <- read.csv("Survival.DSS.PFI.BCAA.csv") BCAA.KM.Pancancer2 <- BCAA.KM.Pancancer %>% filter(Type=="DSS") %>% mutate(Label = if_else(KM.pvalue > 0.05,"p>0.05", if_else(HR < 1,"Protective","Risky"))) %>% #mutate(ENSEMBL=factor(ENSEMBL,levels = rev(c("AASS",Gene.Symble)))) %>% mutate(Label=factor(Label,levels = c("Protective","Risky","p>0.05"))) %>% filter(ENSEMBL!="BCAA") %>% mutate(ENSEMBL=factor(ENSEMBL,levels = rev(sort(unique(.$ENSEMBL))))) #mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) p1 <- ggplot()+ geom_tile(data = BCAA.KM.Pancancer2, aes(y=ENSEMBL,x=Cancer,fill=Label), color="white")+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue>0.05), aes(y=ENSEMBL,x=Cancer,size=-log10(KM.pvalue)), shape=4,color="black")+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue<= 0.05), aes(y=ENSEMBL,x=Cancer,size=-log10(KM.pvalue)), shape=1,color="black")+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(KM p)",fill="Type")+ scale_fill_manual(values = c("#4DBBD5FF","#E5086A","grey90")) ggsave(p1,filename = "Survival.BCAA.DSS.KM.pdf",height = 4,width = 8) p1 <- ggplot()+ geom_point(data = BCAA.KM.Pancancer2, aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(KM.pvalue)), shape=21,color="white",stroke=1)+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue<= 0.05), aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(KM.pvalue)), shape=21,color="black",stroke=1.2)+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(KM p)",fill="Type")+ scale_fill_manual(values = c("#0094FF","#E5086A","#CFB99E")) ggsave(p1,filename = "Survival.BCAA.DSS.KM.pdf",height = 5,width = 4) BCAA.KM.Pancancer3 <- BCAA.KM.Pancancer %>% filter(Type=="DSS") %>% mutate(Label = if_else(HR.pvalue > 0.05,"p>0.05", if_else(HR < 1,"Protective","Risky"))) %>% #mutate(ENSEMBL=factor(ENSEMBL,levels = rev(c("AASS",Gene.Symble)))) %>% mutate(Label=factor(Label,levels = c("Protective","Risky","p>0.05"))) %>% filter(ENSEMBL!="BCAA") %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) p2 <- ggplot()+ geom_point(data = BCAA.KM.Pancancer3, aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(HR.pvalue)), shape=21,color="white",stroke=1)+ geom_point(data = BCAA.KM.Pancancer3 %>% filter(HR.pvalue<= 0.05), aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(HR.pvalue)), shape=21,color="black",stroke=1.2)+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(COX p)",fill="Type")+ scale_fill_manual(values = c("#469990","Orange","grey")) ggsave(p2,filename = "Survival.BCAA.DSS.COX.pdf",height = 5,width = 5) #### Figure => PFI #### library(tidyverse) library(RColorBrewer) BCAA.KM.Pancancer <- read.csv("Survival.DSS.PFI.BCAA.csv") BCAA.KM.Pancancer2 <- BCAA.KM.Pancancer %>% filter(Type=="PFI") %>% mutate(Label = if_else(KM.pvalue > 0.05,"p>0.05", if_else(HR < 1,"Protective","Risky"))) %>% #mutate(ENSEMBL=factor(ENSEMBL,levels = rev(c("AASS",Gene.Symble)))) %>% mutate(Label=factor(Label,levels = c("Protective","Risky","p>0.05"))) %>% filter(ENSEMBL!="BCAA") %>% mutate(ENSEMBL=factor(ENSEMBL,levels = rev(sort(unique(.$ENSEMBL))))) #mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) p1 <- ggplot()+ geom_tile(data = BCAA.KM.Pancancer2, aes(y=ENSEMBL,x=Cancer,fill=Label), color="white")+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue>0.05), aes(y=ENSEMBL,x=Cancer,size=-log10(KM.pvalue)), shape=4,color="black")+ geom_point(data = BCAA.KM.Pancancer2 %>% filter(KM.pvalue<= 0.05), aes(y=ENSEMBL,x=Cancer,size=-log10(KM.pvalue)), shape=1,color="black")+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(KM p)",fill="Type")+ scale_fill_manual(values = c("#4DBBD5FF","#E5086A","grey90")) ggsave(p1,filename = "Survival.BCAA.PFI.KM.pdf",height = 4,width = 8) BCAA.KM.Pancancer3 <- BCAA.KM.Pancancer %>% filter(Type=="PFI") %>% mutate(Label = if_else(HR.pvalue > 0.05,"p>0.05", if_else(HR < 1,"Protective","Risky"))) %>% #mutate(ENSEMBL=factor(ENSEMBL,levels = rev(c("AASS",Gene.Symble)))) %>% mutate(Label=factor(Label,levels = c("Protective","Risky","p>0.05"))) %>% filter(ENSEMBL!="BCAA") %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) p2 <- ggplot()+ geom_point(data = BCAA.KM.Pancancer3, aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(HR.pvalue)), shape=21,color="white",stroke=1)+ geom_point(data = BCAA.KM.Pancancer3 %>% filter(HR.pvalue<= 0.05), aes(y=Cancer,x=ENSEMBL,fill=Label,size=-log10(HR.pvalue)), shape=21,color="black",stroke=1.2)+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="-log10(COX p)",fill="Type")+ scale_fill_manual(values = c("#469990","Orange","grey")) ggsave(p2,filename = "Survival.BCAA.PFI.COX.pdf",height = 5,width = 5) ######## Methylation ############ BCAA.SYMBOLs <- read.csv("BCAA.ID.csv") library(readxl) # methylation <- read_xlsx("GSCA/ExpressionAndMethylationTable.xlsx") library(tidyverse) library(magrittr) PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) GPL <- read.csv("../../Clinical.XENA/Methylation450K.GPL13534-11288.txt",sep = "\t",header = T,comment.char = "#") GPL <- GPL %>% dplyr::select(ID, UCSC_RefGene_Name, UCSC_RefGene_Group, Regulatory_Feature_Group, Relation_to_UCSC_CpG_Island) BCAA.GPL <- data.frame() for (i in 1:nrow(GPL)) { print(i) SYMBOL=unique(unlist(str_split(GPL$UCSC_RefGene_Name[i],";"))) if (length(SYMBOL) >= 1) { for (GENE in SYMBOL) { if (GENE %in% BCAA.SYMBOLs$SYMBOL) { BCAA.GPL <- data.frame(ID=GPL$ID[i], UCSC_RefGene_Name=GENE, UCSC_RefGene_Group=paste(unique(sort(unlist(str_split(GPL$UCSC_RefGene_Group[i],";")))),collapse = ";"), Regulatory_Feature_Group=GPL$Regulatory_Feature_Group[i], Relation_to_UCSC_CpG_Island=GPL$Relation_to_UCSC_CpG_Island[i]) %>% rbind.data.frame(BCAA.GPL) } } } } BCAA.GPL2 <- BCAA.GPL %>% filter(str_detect(ID,"cg")) %>% arrange(UCSC_RefGene_Name) write.csv(BCAA.GPL2,file = "Methylation450K.BCAA.GPL.ID.csv",row.names = F) #### Spearman => Data preparation #### library(tidyverse) BCAA.GPL2 <- read.csv(file = "Methylation450K.BCAA.GPL.ID.csv") #PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) PROJECTS <- unique(PhenoData$Project.ID) %>% str_remove_all("TCGA-") => Datrfor (project in PROJECTS) { print(project) Cancer=paste("TCGA-",project,sep = "") mid.Pheno <- PhenoData %>% filter(Project.ID == Cancer) %>% dplyr::select(Project.ID,Sample.ID,Sample.Type,SampleType) File.path = paste("../TotalRNA.Microbiome/Clinical.XENA/",project,"/",sep = "") #File.path = paste(project,"/",sep = "") File.name = paste("TCGA-",project,".methylation450.tsv",sep = "") Data <- read.csv(file = file.path(File.path,File.name),sep = "\t",check.names = F,row.names = 1) Data2 <- Data %>% rownames_to_column("ID") %>% #filter(ID %in% BCAA.GPL2$ID) %>% gather(SampleID,Methylation,-ID) %>% na.omit() %>% filter(SampleID %in% mid.Pheno$Sample.ID) %>% merge(mid.Pheno,by.x="Sample.ID",by.y="Sample.ID") %>% merge(BCAA.GPL2,by="ID") Output.name=paste("XENA.BCAA.Methylation450K.",project,".csv",sep = "") write.csv(Data2,file=Output.name) } #### Spearman => Calculate #### library(tidyverse) BCAA.GPL2 <- read.csv(file = "Methylation450K.BCAA.GPL.ID.csv") PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) %>% mutate(Project.ID=str_remove(Project.ID,"TCGA-")) BCAA.ID <- read.csv("BCAA.ID.csv") #%>% filter(ENSEMBL!="ENSG00000285069") library(survival) library(survminer) Methylation.Wilcox <- data.frame() Methylation.Exp.Spearman <- data.frame() Methylation.Survival <- data.frame() files <- list.files(path = "Methylation450K",pattern = "txt$") for (file in files) { project = (str_split(file,"\\.") %>% unlist())[2] print(project) methylated.data <- read.csv(file.path("Methylation450K",file),sep = "\t",row.names = 1,check.names = F) #methylated.data[is.na(methylated.data)] = 0 methylated.data <- methylated.data %>% t() %>% data.frame(check.names = F) NA.count <- sapply(methylated.data, function(x) sum(is.na(x))) methylated.data2 <- methylated.data[,NA.count < nrow(methylated.data)] %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") if (any(str_detect(methylated.data2$Sample.ID,"-??A$"))) { mid.Pheno <- PhenoData %>% filter(Project.ID==project) %>% dplyr::select(Sample.ID,SampleType) %>% merge(methylated.data2,by="Sample.ID") %>% mutate(SampleID=substr(Sample.ID,1,15)) }else{ mid.Pheno <- PhenoData %>% filter(Project.ID==project) %>% dplyr::select(Sample.ID,SampleType) %>% mutate(SampleID=substr(Sample.ID,1,15)) %>% merge(methylated.data2,by.x="SampleID",by.y="Sample.ID") } #### Spearman => Methylation + Expression #### Exp.filename <- paste("TCGA-",project,".mRNA.TPM.csv",sep = "") Exp <- read.csv(file = file.path("../../mRNA.PanCancer.Exp/",Exp.filename),row.names = 1,check.names = F) Exp <- log2(Exp+1) BCAA.Exp <- Exp[BCAA.ID$ENSEMBL,] %>% data.frame(check.names = F) %>% rownames_to_column("ENSEMBL") %>% merge(BCAA.ID,by="ENSEMBL") %>% dplyr::select(-ENSEMBL,-ENTREZ) %>% column_to_rownames("SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% merge(mid.Pheno,by="Sample.ID") for (symbol in unique(BCAA.GPL2$UCSC_RefGene_Name)) { mid.cg <- BCAA.GPL2 %>% filter(UCSC_RefGene_Name==symbol) if (symbol == "GATSL2") { symbol="CASTOR2" }else if(symbol == "GATSL3"){ symbol="CASTOR1" } for (cgid in mid.cg$ID) { if (cgid %in% colnames(BCAA.Exp)) { test1 <- cor.test(BCAA.Exp[[cgid]],BCAA.Exp[[symbol]], method = "spearman", exact = F) Methylation.Exp.Spearman <- data.frame(SYMBOL=symbol, ID=cgid, Rho=test1$estimate, Pvalue=test1$p.value, Cancer=project, Count=length(BCAA.Exp[[cgid]])) %>% rbind.data.frame(Methylation.Exp.Spearman) } } } #### Survival => Methylation #### TUMOR <- mid.Pheno %>% filter(SampleType=="Tumor") %>% mutate(SampleID=substr(Sample.ID,1,15)) Sur.filename <- paste("XENA.TCGA.",project,"_survival.txt",sep = "") Sur.Data <- read.csv(file.path("../../Clinical.XENA/",Sur.filename),row.names = 1,sep = "\t",check.names = F) %>% rownames_to_column("SampleID") %>% merge(TUMOR,by="SampleID") for (id in colnames(TUMOR)[colnames(TUMOR) %>% str_detect("cg")]) { Sur.Data2 <- Sur.Data %>% dplyr::select(c("OS","OS.time",id)) %>% na.omit() Sur.Data2$group=ifelse(Sur.Data2[[id]] > median(Sur.Data2[[id]]),"high","low") if(length(table(Sur.Data2$group))==1) print("Only one group") index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","group",sep = "")), data = Sur.Data2) pValue=1-pchisq(diff$chisq,df=1) cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ",id,sep = "")), data = Sur.Data2) coxSummary = summary(cox) Methylation.Survival=rbind(Methylation.Survival, data.frame(ID=id, KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project)) } CTRL <- mid.Pheno %>% filter(SampleType=="Normal") #### Wilcox #### print(nrow(CTRL)) if (nrow(CTRL) >= 10 & nrow(TUMOR) >= 10) { Tumor.Samples <- mid.Pheno %>% filter(SampleType=="Tumor") Ctrl.Samples <- mid.Pheno %>% filter(SampleType=="Normal") for (cg.id in colnames(methylated.data2[,2:ncol(methylated.data2)])) { test <- wilcox.test(Tumor.Samples[[cg.id]], Ctrl.Samples[[cg.id]], exact = F, alternative = "two.sided") Methylation.Wilcox <- data.frame(ID=cg.id, Wilcox.pvalue=test$p.value, Mean.Ctrl = mean(na.omit(Ctrl.Samples[[cg.id]])), Median.Ctrl = median(na.omit(Ctrl.Samples[[cg.id]])), Mean.Tumor = mean(na.omit(Tumor.Samples[[cg.id]])), Median.Tumor = median(na.omit(Tumor.Samples[[cg.id]])), Cancer=project) %>% rbind.data.frame(Methylation.Wilcox) } } } #Methylation.BCAA <- Methylation.Exp.Spearman %>% remove_rownames() %>% # #merge(Methylation.Wilcox,by=c("ID","Cancer")) %>% # merge(Methylation.Survival,by=c("ID","Cancer")) %>% # merge(BCAA.GPL2,by="ID") write.csv(Methylation.Wilcox,file = "Methylation.BCAA.Wilcox.csv",row.names = F) write.csv(Methylation.Exp.Spearman,file = "Methylation.BCAA.Exp.Spearman.csv",row.names = F) write.csv(Methylation.Survival,file = "Methylation.BCAA.Survival.csv",row.names = F) #### Figure => All methylation sites #### #Methylation.BCAA <- read.csv(file = "Methylation.BCAA.Survival.Wilcox.csv") Methylation.Exp.Spearman <- read.csv(file = "Methylation.BCAA.Exp.Spearman.csv") read.csv(file = "Methylation.BCAA.Exp.Spearman.csv") %>% filter(Pvalue < 0.05) %>% write.csv("TableS3.csv",row.names = F) Methylation.Survival <- read.csv(file = "Methylation.BCAA.Survival.csv") #Methylation.Survival2 <- round(Methylation.Survival,1e20) Methylation.BCAA2 <- Methylation.Exp.Spearman %>% dplyr::select(ID,Cancer,Rho) %>% #pivot_longer(-gene) spread(Cancer,Rho,fill=0) %>% remove_rownames() %>% column_to_rownames("ID") library(ComplexHeatmap) library(circlize) library(tidyverse) library(ggtree) library(aplot) library(ggh4x) ## heatmap heatmap <- ggplot(Methylation.Exp.Spearman,aes(x=Cancer,y=ID,fill=Rho))+ geom_tile()+ scale_fill_gradient2(mid="white",low="#313695",high="#A50026")+ guides(fill=guide_colorbar(direction = "vertical", reverse = F,barwidth = unit(.6, "cm"), barheight = unit(3,"cm")))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="Spearman'CC") ## cancer Cluster cancerorder <- hclust(dist(t(Methylation.BCAA2))) phc <- hclust(dist(t(Methylation.BCAA2))) %>% ggtree() + layout_dendrogram()+ theme_void() ## methylation sites Cluster phr <- hclust(dist(Methylation.BCAA2)) %>% ggtree(layout="rectangular",branch.length="none")+ theme_void() ## methylation type library(RColorBrewer) Cg.group <- BCAA.GPL2 %>% ggplot(aes(x=1,y=ID,fill=UCSC_RefGene_Group))+ geom_col(width=1,color="black",size=0.01)+ theme_void()+ scale_fill_manual(values = c(brewer.pal(9,"Set1"),brewer.pal(12,"Set3"))) Cg.cpg <- BCAA.GPL2 %>% ggplot(aes(x=1,y=ID,fill=Relation_to_UCSC_CpG_Island))+ geom_col(width=1,color="black",size=0.01)+ theme_void()+ scale_fill_manual(values = c("white",brewer.pal(8,"Dark2")[c(1,2,3,4,6)])) ## methylation genes Colors17 <- c("#949483","#F47B7B","#9F1F5C","#EF9020", "#00AF3E","#85B7E2","#29245C","#FFD616", "#E5352B","#E990AB","#0081B4","#96CBB3", "#91BE3E","#39A6DD","#EB0973","#DDE2E0", "#333C41","#9013FE","#FFC208","#1FADC5", "#2B697A") Cg.genes <- Methylation.Exp.Spearman %>% dplyr::select(ID,SYMBOL) %>% unique() %>% mutate(SYMBOL=factor(SYMBOL,levels = BCAA.ID$SYMBOL)) %>% ggplot(aes(x=1,y=ID,fill=SYMBOL))+ geom_col(width=1,color="black",size=0.01)+ theme_void()+ scale_fill_manual(values = Colors17) ## methylation survival Cg.survival <- Methylation.Survival %>% mutate(Type = if_else(KM.pvalue <= 0.05 | HR.pvalue <= 0.05, if_else(HR > 1,"Risky","Protective"),"p>0.05")) %>% mutate(Type=factor(Type,levels = c("Protective","Risky","p>0.05"))) %>% mutate(Cancer=factor(Cancer,levels = colnames(Methylation.BCAA2)[cancerorder$order])) %>% ggplot(aes(x=Cancer,y=ID,fill=Type))+ geom_tile()+ # ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_blank(), axis.ticks.y=element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="")+ scale_fill_manual(values = c(brewer.pal(8,"Set1")[1],brewer.pal(8,"Set1")[2],"white")) ## Final BCAA.Methylation.All <- heatmap %>% #insert_top(cell_type,height=0.05) %>% aplot::insert_top(phc,height=0.05) %>% #insert_left(gene_type,width = 0.06) %>% aplot::insert_left(phr,width = 0.1) %>% aplot::insert_right(Cg.genes,width = 0.05) %>% aplot::insert_right(Cg.group,width = 0.05) %>% aplot::insert_right(Cg.cpg,width = 0.05) %>% aplot::insert_right(Cg.survival,width = 1) ggsave(BCAA.Methylation.All,filename = "Methylation.BCAA.Figure.pdf",height = 8,width = 12) ggsave(BCAA.Methylation.All,filename = "Methylation.BCAA.Figure.Legend.pdf",height = 16,width = 12) #### Figure => Representative #### Methylation.Exp.Spearman <- read.csv(file = "Methylation.BCAA.Exp.Spearman.csv") %>% mutate(Cancer=paste(Cancer,"(n=",Count,")",sep = "")) %>% dplyr::select(-Count) Methylation.Survival <- read.csv(file = "Methylation.BCAA.Survival.csv") Methylation.Wilcox <- read.csv(file = "Methylation.BCAA.Wilcox.csv") BCAA.GPL2 <- read.csv("Methylation450K.BCAA.GPL.ID.csv") CgID <- BCAA.GPL2 %>% filter(!str_detect(UCSC_RefGene_Group,"Body")) Methylation.Survival2 <- Methylation.Survival %>% filter(HR < 1) %>% filter(KM.pvalue <= 0.05) Methylation.Exp <- Methylation.Exp.Spearman %>% filter(Pvalue <= 0.05) %>% filter(Rho <= 0) %>% #filter(ID %in% CgID$ID) %>% group_by(ID) %>% summarise(Count=n()) %>% ungroup() %>% arrange(desc(Count)) %>% merge(Methylation.Exp.Spearman,by="ID") %>% dplyr::select(ID,Count,SYMBOL) %>% unique() %>% data.frame(check.names = F) %>% group_by(SYMBOL) %>% dplyr::top_n(2,Count) %>% merge(Methylation.Exp.Spearman,by=c("ID","SYMBOL")) %>% mutate(SYMBOL=factor(SYMBOL,levels = BCAA.ID$SYMBOL)) %>% arrange(SYMBOL,ID) %>% mutate(ID=factor(ID,levels = rev(unique(.$ID)))) Figure.Body <- ggplot()+ geom_point(data=Methylation.Exp, aes(x=Cancer,y=ID,fill=Rho,size=-log10(Pvalue)), shape=21,color="white",stroke=1) + geom_point(data=Methylation.Exp %>% filter(Pvalue<=0.05), aes(x=Cancer,y=ID,fill=Rho,size=-log10(Pvalue)), shape=21,color="black",stroke=1)+ scale_size_continuous(range = c(1,5))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), #axis.text.y = element_blank(), #axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="Spearman'CC")+ scale_fill_gradient2(mid="white",high="#A50026",low="#313695") Figure.R2 <- Methylation.Exp %>% mutate(Label=if_else(Pvalue > 0.05,"P>0.05", if_else(Rho > 0,"Positive","Negative","P>0.05"))) %>% group_by(ID,Label) %>% summarise(Count=n()) %>% spread(Label,Count,fill=0) %>% mutate(ID=factor(ID,levels = levels(Methylation.Exp$ID))) %>% arrange(ID) %>% gather(Type,Count,-ID) %>% mutate(ID=factor(ID,levels = levels(Methylation.Exp$ID))) %>% mutate(Type=factor(Type,levels = rev(c("Positive","Negative","P>0.05")))) %>% ggplot(aes(x=Count,y=ID,fill=Type))+ geom_col(color="black")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="Cancer count",y="",fill="Correlation")+ scale_fill_manual(values = rev(c("#E7298A","#1B9E77","grey"))) Colors17 <- c("#949483","#F47B7B","#9F1F5C","#EF9020", "#00AF3E","#85B7E2","#29245C","#FFD616", "#E5352B","#E990AB","#0081B4","#96CBB3", "#91BE3E","#39A6DD","#EB0973","#DDE2E0", "#333C41","#9013FE","#FFC208","#1FADC5", "#2B697A") Figure.R1 <- Methylation.Exp %>% dplyr::select(ID,SYMBOL) %>% unique() %>% mutate(SYMBOL=factor(SYMBOL,levels = BCAA.ID$SYMBOL)) %>% ggplot(aes(x=1,y=ID,fill=SYMBOL))+ geom_col(color="black")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_blank(), axis.text.y = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="Symbol")+ scale_fill_manual(values = Colors17) library(RColorBrewer) Cg.group <- BCAA.GPL2 %>% ggplot(aes(x=1,y=ID,fill=UCSC_RefGene_Group))+ geom_col(width=1,color="black",size=0.01)+ theme_void()+ scale_fill_manual(values = c(brewer.pal(9,"Set1"),brewer.pal(12,"Set3"))) Cg.cpg <- BCAA.GPL2 %>% ggplot(aes(x=1,y=ID,fill=Relation_to_UCSC_CpG_Island))+ geom_col(width=1,color="black",size=0.01)+ theme_void()+ scale_fill_manual(values = c("white",brewer.pal(8,"Dark2")[c(1,2,3,4,6)])) Figure.F <- Figure.Body %>% aplot::insert_right(Figure.R1,width = 0.05) %>% aplot::insert_right(Cg.group,width = 0.05) %>% aplot::insert_right(Cg.cpg,width = 0.05) %>% aplot::insert_right(Figure.R2,width = 0.2) ggsave(Figure.F,filename = "Methylation.BCAA.Figure.Representative.pdf",height = 8,width = 10) ggsave(Figure.F,filename = "Methylation.BCAA.Figure.Representative.Legend.pdf",height = 16,width = 10) ######## ATAC ######## setwd("K:/TCGA/Anlysis/BCAA") BCAA.ID <- read.csv("BCAA.ID.csv") setwd("K:/TCGA/Cancer.ATAC") Sequencing.Data <- read.csv("Cancer.ATAC.Metadata.csv") ATAC.BCAA <- data.frame() for (project in unique(Sequencing.Data$cohort)) { Anno <- read.csv(paste("ATAC.",project,".PeakChIPSeeker.csv",sep = "")) SNHGs.Peak <- Anno %>% filter(SYMBOL %in% BCAA.ID$SYMBOL) %>% dplyr::select(Peak_name,SYMBOL) %>% unique() FPKM.ATAC <- read.csv(paste("ATAC.",project,".FPKM.csv",sep = ""),row.names = 1) ATAC.BCAA <- FPKM.ATAC[SNHGs.Peak$Peak_name,] %>% data.frame() %>% rownames_to_column("Peak_name") %>% gather(Sample,FPKM,-Peak_name) %>% merge(.,SNHGs.Peak,by="Peak_name") %>% mutate(Cancer = project) %>% rbind.data.frame(ATAC.BCAA) } setwd("K:/TCGA/Anlysis/BCAA") write.csv(ATAC.BCAA,file = "BCAA.ATAC.FPKM.csv",row.names = F) #### Data preparation #### setwd("K:/TCGA/Cancer.ATAC") Sequencing.Data <- read.csv("Cancer.ATAC.Metadata.csv") Cancer.ATAC <- data.frame() for (name in BCAA.ID$SYMBOL) { Cancer.ATAC <- rbind(Cancer.ATAC,data.frame(SYMBOL=name,Cancer=unique(Sequencing.Data$cohort),MeanFPKM=0)) } setwd("K:/TCGA/Anlysis/BCAA") ATAC.BCAA <- read.csv(file = "BCAA.ATAC.FPKM.csv") write.csv(ATAC.BCAA,file = "ATAC.BCAA.FPKM.csv",row.names = F) #### Figure #### ATAC.BCAA <- read.csv(file = "ATAC.BCAA.FPKM.csv") ATAC.BCAA.FPKM <- ATAC.BCAA %>% group_by(SYMBOL,Cancer) %>% summarise(MeanFPKM = mean(FPKM), SDFPKM=sd(FPKM), CVFPKM=round(MeanFPKM/SDFPKM,4)*100) %>% ungroup() %>% mutate(SYMBOL=factor(SYMBOL,levels = rev(BCAA.ID$SYMBOL))) write.csv(ATAC.BCAA.FPKM,"ATAC.BCAA.FPKM.MeanCV.csv",row.names = F) library(RColorBrewer) p1 <- ggplot(data=ATAC.BCAA.FPKM,aes(x=Cancer, y=SYMBOL, fill=MeanFPKM, size=MeanFPKM))+ geom_point(shape=22)+ scale_size_continuous(range = c(1,5))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="Mean of FPKM",fill="CV of FPKM")+ scale_fill_gradientn(colors = brewer.pal(11,"RdYlBu")[4:1]) ggsave(p1,filename = "ATAC.BCAA.FPKM.MeanCV.pdf",height = 3,width = 6) #### BCAT1 #### ATAC.BCAA <- read.csv(file = "BCAA.ATAC.FPKM.csv") BCAT1.ATAC <- ATAC.BCAA %>% filter(SYMBOL=="BCAT1") %>% mutate(Cancer=factor(Cancer,levels = rev(unique(sort(.$Cancer))))) Immune.Col <- c("#293462","#a64942","#fe5f55","#fff1c1", "#5bd1d7","#348498","#004d61","#ff502f", "#e41749","#f5587b","#ff8a5c","#fff591", "#001871","#ff585d","#ffb549","#41b6e6", "#515bd4","#8134af","#dd2a7b","#feda77", "#96ceb4","#ffeead","#d9534f","#ffad60", "#05445c","#f2317f","#5c4f74","#040000", "#de4307","#f29c2b","#f6d04d","#8bc24c", "#fef4a9","#3b9a9c","#4bc2c5","#78fee0", "#fad3cf","#a696c8","#2470a0","#c50d66", "#f07810","#eec60a","#dd7777","#1687a7", "#014955","#dd0a35","#ED5485","#FFE869", "#bc8420","#F0B775","#D25565","#2E94B9") p3<-ggplot(BCAT1.ATAC ,aes(x=FPKM,y=reorder(Cancer,FPKM),fill=Cancer))+ #geom_violin()+ geom_boxplot(color="black")+ ggthemes::theme_few()+ theme(legend.position = "none", #axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,color = "black"), axis.text.y = element_text(size=9,color = "black"), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_x_sqrt()+ scale_x_continuous(breaks = c(0,1,5,10,20,40,60,80), labels = c(0,1,5,10,20,40,60,80), trans = "sqrt")+ scale_fill_manual(values = Immune.Col)+ labs(y="",x="FPKM of BCAT1") ggsave(p3,filename = "ATAC.BCAA.BCAT1.FPKM.pdf",height = 5,width = 2.3) #### BCAT2 #### ATAC.BCAA <- read.csv(file = "BCAA.ATAC.FPKM.csv") ATAC.BCAA %>% filter(SYMBOL=="BCAT2") %>% group_by(Cancer) %>% top_n(2,FPKM) %>% View() BCAT2.ATAC <- ATAC.BCAA %>% filter(SYMBOL=="BCAT2") %>% mutate(Cancer=factor(Cancer,levels = rev(unique(sort(.$Cancer))))) Immune.Col <- c("#293462","#a64942","#fe5f55","#fff1c1", "#5bd1d7","#348498","#004d61","#ff502f", "#e41749","#f5587b","#ff8a5c","#fff591", "#001871","#ff585d","#ffb549","#41b6e6", "#515bd4","#8134af","#dd2a7b","#feda77", "#96ceb4","#ffeead","#d9534f","#ffad60", "#05445c","#f2317f","#5c4f74","#040000", "#de4307","#f29c2b","#f6d04d","#8bc24c", "#fef4a9","#3b9a9c","#4bc2c5","#78fee0", "#fad3cf","#a696c8","#2470a0","#c50d66", "#f07810","#eec60a","#dd7777","#1687a7", "#014955","#dd0a35","#ED5485","#FFE869", "#bc8420","#F0B775","#D25565","#2E94B9") p3<-ggplot(BCAT2.ATAC ,aes(x=FPKM,y=reorder(Cancer,FPKM),fill=Cancer))+ #geom_violin()+ geom_boxplot(color="black")+ ggthemes::theme_few()+ theme(legend.position = "none", #axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,color = "black"), axis.text.y = element_text(size=9,color = "black"), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_x_sqrt()+ scale_x_continuous(breaks = c(0,1,5,10,20,40,60,80), labels = c(0,1,5,10,20,40,60,80), trans = "sqrt")+ scale_fill_manual(values = Immune.Col)+ labs(y="",x="FPKM of BCAT2") ggsave(p3,filename = "ATAC.BCAA.BCAT2.FPKM.pdf",height = 5,width = 2.3) ######## NMF ############ #### Cluster => TPM => NMF #### library(NMF) library(tidyverse) dir.create("NMF.BCAA") BCAA.ID <- read.csv("BCAA.ID.csv") #PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% PhenoData <- read.csv("../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) %>% mutate(Project.ID=str_remove(Project.ID,"TCGA-")) %>% filter(SampleType == "Tumor") for (project in unique(PhenoData$Project.ID)) { print(project) mid.Pheno <- PhenoData %>% filter(Project.ID == project) #filepath = "../../mRNA.PanCancer.Exp/" filepath = "../mRNA.PanCancer.Exp/" filename = paste("TCGA-",project,".mRNA.TPM.csv",sep = "") data <- read.csv(file.path(filepath,filename),check.names = F,row.names = 1) BCAA.Data <- data[BCAA.ID$ENSEMBL,] %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% filter(Sample.ID %in% mid.Pheno$Sample.ID) %>% column_to_rownames("Sample.ID") %>% t() ranks <- 2:10 seed <- 20230206 figure.name <- paste(project,".NMF.2-10.pdf",sep = "") file.name <- paste(project,".NMF.2-10.rds",sep = "") result = nmf(BCAA.Data, ranks, method="brunet", nrun=5, seed =seed) saveRDS(result,file = file.path("NMF.BCAA",file.name)) pdf(file = figure.name) plot(result) dev.off() } #### Cluster => Confirm #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) files <- list.files(path = "NMF.BCAA/",pattern = "rds$") NMF.Rank <- data.frame() for (file in files) { data <- readRDS(file.path("NMF.BCAA/",file)) NMF.Rank <- data$measures[,c("cophenetic","rank")] %>% data.frame(check.names = F) %>% mutate(Cancer=file %>% str_remove_all("\\..*")) %>% rbind.data.frame(NMF.Rank) } write.csv(NMF.Rank,file = "NMF.BCAA/NMF.BCAA.Cophenetic.csv",row.names = F) ggplot(NMF.Rank,aes(rank,cophenetic,color=Cancer))+ geom_point(size=2)+ geom_line()+ facet_wrap(~Cancer,ncol = 6)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) ######## miRNA => Network Pearson ######## #### multiMiR #### library(multiMiR) library(plyr) BCAA.ID <- read.csv("BCAA.ID.csv") multiMiR.Data <- data.frame() for (symbol in BCAA.ID$SYMBOL) { miRNA.Target <- multiMiR::get_multimir(target = symbol,table="all",org = 'hsa',summary = TRUE) multiMiR.Data <- miRNA.Target@data %>% rbind.fill(multiMiR.Data) } write.csv(multiMiR.Data,file = "miRNA.BCAA.multiMiR.csv") #### TCGA => miRNA + mRNA => Correlation #### library(miRBaseVersions.db) library(vegan) Inter.miRNA.mRNA <- read.csv(file = "miRNA.BCAA.multiMiR.csv",row.names = 1) Inter.miRNA.mRNA <- Inter.miRNA.mRNA[,c(2:6)] %>% data.frame() %>% unique() BCAA.ID <- read.csv("BCAA.ID.csv") dirs <- list.dirs("../../Clinical.XENA/") dirs <- dirs[str_detect(dirs,"//")] miRNA.mRNA.Pearson <- data.frame() for (dir in dirs) { project <- dir %>% str_remove(".*\\/") print(project) mRNA.Filename <- paste("TCGA-",project,".mRNA.TPM.csv",sep = "") mRNA.Exp <- read.csv(file.path("../../mRNA.PanCancer.Exp/",mRNA.Filename),row.names = 1,check.names = F) mRNA.Exp <- log2(mRNA.Exp+1) BCAA.Exp <- mRNA.Exp[BCAA.ID$ENSEMBL,] %>% data.frame(check.names = F) %>% rownames_to_column("ENSEMBL") %>% merge(BCAA.ID,by="ENSEMBL") %>% dplyr::select(-ENSEMBL,-ENTREZ) %>% remove_rownames() %>% column_to_rownames("SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") miRNA.Filename <- paste("TCGA-",project,".mature.miRNA.RPM.csv",sep = "") miRNA.Exp <- read.csv(file.path("../../miRNA.Mature.PanCancer.Exp/",miRNA.Filename), check.names = F,row.names = 1) miRNA.mRNA.Exp <- miRNA.Exp %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% merge(BCAA.Exp,by="Sample.ID") if (nrow(miRNA.mRNA.Exp) > 0) { for (gene in BCAA.ID$SYMBOL) { mid.gene.miRNA <- Inter.miRNA.mRNA %>% filter(target_symbol==gene) if (nrow(mid.gene.miRNA) > 0) { print("Samples overlapped") for (miRNA.id in mid.gene.miRNA$mature_mirna_id) { if (c(miRNA.id) %in% colnames(miRNA.mRNA.Exp)) { test <- cor.test(miRNA.mRNA.Exp[[miRNA.id]], miRNA.mRNA.Exp[[gene]], method = "pearson", exact = F, alternative = "two.sided") miRNA.mRNA.Pearson <- data.frame(miRNA=miRNA.id, Target=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(miRNA.mRNA.Pearson) } } }else( print("No overlap") ) } } } miRNA.mRNA.Pearson2 <- miRNA.mRNA.Pearson %>% na.omit() %>% mutate(Label=if_else(Pvalue>0.05,"p>0.05", if_else(Rho>0,"Positive","Negative"))) write.csv(miRNA.mRNA.Pearson2,file = "miRNA.TCGA.miRNA.mRNA.Pearson.csv",row.names = F) #### Figure #### miRNA.mRNA.Pearson2 <- read.csv(file = "miRNA.TCGA.miRNA.mRNA.Pearson.csv") library(plyr) miRNA.mRNA.Pearson3 <- miRNA.mRNA.Pearson2 %>% data.frame(check.names = F) %>% group_by(Target,miRNA,Label) %>% dplyr::summarise(Count=n()) %>% ungroup() %>% filter(Label=="Negative") %>% arrange(desc(Count)) %>% dplyr::top_n(30,Count) %>% dplyr::select(Target,miRNA) %>% arrange(Target,miRNA) %>% unique() %>% merge(miRNA.mRNA.Pearson2,by=c("Target","miRNA")) %>% mutate(Yaxis=paste(miRNA,Target,sep = "->")) %>% arrange(Target,miRNA,Cancer) %>% unique() %>% mutate(Yaxis=factor(Yaxis,levels = rev(unique(.$Yaxis)))) write.csv(miRNA.mRNA.Pearson3,file = "miRNA.TCGA.miRNA.mRNA.Pearson.Data.csv",row.names = F) library(RColorBrewer) Figure.Body <- ggplot()+ geom_point(data=miRNA.mRNA.Pearson3, aes(x=Cancer,y=Yaxis,fill=Rho,size=-log10(Pvalue)), shape=22,color="white",stroke=1)+ geom_point(data=miRNA.mRNA.Pearson3 %>% filter(Pvalue <= 0.05), aes(x=Cancer,y=Yaxis,fill=Rho,size=-log10(Pvalue)), shape=22,color="black",stroke=1)+ scale_size_continuous(range = c(2,5))+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",size="p value",fill="Pearson'CC")+ scale_fill_gradient2(high = "#BC3C29FF", mid = "white", low = "#0072B5FF",midpoint = 0) #"#0072B5FF","#BC3C29FF" Figure.R1 <- miRNA.mRNA.Pearson3 %>% dplyr::select(Yaxis,Target) %>% unique() %>% ggplot(aes(x=1,y=Yaxis,fill=Target))+ geom_col(width = 1,color="black")+ scale_fill_manual("Symbol",values = brewer.pal(12,"Paired"))+ theme_void() Figure.R2 <- miRNA.mRNA.Pearson3 %>% group_by(Target,miRNA,Label) %>% dplyr::summarise(Count=n()) %>% #arrange(desc(Count)) ungroup() %>% mutate(Yaxis=paste(miRNA,Target,sep = "->")) %>% dplyr::select(-Target,-miRNA) %>% spread(Label,Count,fill=0) %>% gather(Correlation,Count,-Yaxis) %>% mutate(Correlation=factor(Correlation, levels = c("p>0.05","Negative","Positive"))) %>% ggplot(aes(x=Count,y=Yaxis,fill=Correlation))+ geom_col(width = 1,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_blank(), axis.ticks.y=element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="No. of cancers",y="")+ scale_fill_manual(values = c("grey90","#66C2A5","#BC3C2999")) Figure.F <- Figure.Body %>% aplot::insert_right(Figure.R1,width = 0.05) %>% aplot::insert_right(Figure.R2,width = 0.3) ggsave(Figure.F,filename = "miRNA.TCGA.miRNA.mRNA.Pearson.pdf",height = 6,width = 9) ggsave(Figure.F,filename = "miRNA.TCGA.miRNA.mRNA.Pearson.Legend.pdf",height = 12,width = 10) ######## BCAA score ssGSEA ####### library(GSVA) library(GSEABase) library(msigdbr) library(clusterProfiler) library(org.Hs.eg.db) library(tidyverse) BCAA.ID <- read.csv("BCAA.ID.csv") BCAA <- list() BCAA[["BCAA"]] <- BCAA.ID$ENSEMBL dir.create("BCAA.ssGSEA") BCAA.ssGSEA <- data.frame() files <- list.files("../../mRNA.PanCancer.Exp/",pattern = "csv$") for (file in files) { project <- file %>% str_remove_all("\\..*") %>% str_remove_all("TCGA-") print(project) data <- read.csv(file.path("../../mRNA.PanCancer.Exp/",file),row.names = 1,check.names = F) data2 <- log2(data+1) BCAA.ssGSEA <- gsva(as.matrix(data2), BCAA, method = "ssgsea", kcdf = "Gaussian") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% mutate(Cancer=project) %>% rbind.data.frame(BCAA.ssGSEA) #write.csv(res.ssgsea,file = paste("HALLMARK.ssGSEA.Pancancer/TCGA-",project,".ssGSEA.csv",sep = "")) } write.csv(BCAA.ssGSEA,"BCAA.ssGSEA/BCAA.ssGSEA.csv",row.names = F) ######## BCAA score Survival #### library(survival) library(survminer) library(tidyverse) library(export) files <- list.files("../../mRNA.PanCancer.Exp/",pattern = "csv$") BCAA.Survival <- data.frame() for (file in files) { project <- file %>% str_remove_all("\\..*") %>% str_remove_all("TCGA-") print(project) #data <- read.csv(file.path("../../mRNA.PanCancer.Exp/",file),row.names = 1,check.names = F) #TPM <- log2(data+1) TPM <- BCAA.ssGSEA %>% filter(Cancer==project) phenotype <- read.csv(paste("../../Clinical.XENA/",project,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") #### survival datasets survival.data <- read.csv(paste("../../Clinical.XENA/",project,"/TCGA-",project,".survival.tsv",sep = ""),sep = "\t",header = T,row.names = 1,check.names = F) %>% rownames_to_column("Sample.ID") Inter.Samples <- intersect(intersect(phenotype$submitter_id.samples,TPM$Sample.ID),survival.data$Sample.ID) #### survival analysis TPM.log.Target <- TPM %>% filter(Sample.ID %in% Inter.Samples) %>% merge(survival.data,by="Sample.ID") res.cut <- surv_cutpoint(TPM.log.Target, time = paste("OS",".time",sep = ""), event = "OS", variables = "BCAA") #res.cat <- surv_categorize(res.cut) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(Surv(OS.time, OS) ~ BCAA, data = dat) index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","BCAA",sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ","BCAA",sep = "")), data = TPM.log.Target) coxSummary = summary(cox) library(ggthemes) p<-ggsurvplot(fit,pval =TRUE, data = dat, surv.median.line = "hv", legend.title = "BCAA", conf.int.style = "step", xlab = "Time in days", #break.time.by = 500, risk.table = "abs_pct", #risk.table.y.text.col = T, #risk.table.y.text = FALSE, legend.labs = c("High", "Low"), #pval = TRUE, conf.int = TRUE, palette = "Set1", #ggtheme = theme.sur, risk.table.y.text.col = T, risk.table.y.text = F, ggtheme = theme.sur) print(p) #dev.off() graph2pdf(file=paste("BCAA.ssGSEA/TCGA-",project,".OS.BCAA.KM.pdf",sep = ''),height = 5,width = 4) BCAA.Survival=rbind(BCAA.Survival, data.frame(ENSEMBL="BCAA", KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project)) } write.csv(BCAA.Survival,file = "BCAA.ssGSEA/BCAA.Survival.csv",row.names = F) #### Figure #### BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") library(survival) library(survminer) library(tidyverse) BCAA.Survival <- read.csv(file = "BCAA.ssGSEA/BCAA.Survival.csv") BCAA.Survival <- BCAA.Survival %>% mutate(Colors=if_else(KM.pvalue >0.05 & HR.pvalue > 0.05,"KM p>0.05 & COX p>0.05", if_else(KM.pvalue >0.05 & HR.pvalue <= 0.05,"KM p>0.05 & COX p≤0.05", if_else(KM.pvalue <=0.05 & HR.pvalue > 0.05,"KM p≤0.05 & COX p>0.05","KM p≤0.05 & COX p≤0.05")))) %>% arrange(HR) %>% mutate(Color=if_else(HR.pvalue > 0.05,"P>=0.05","P<=0.05")) BCAA.Survival$Cancer <- factor(BCAA.Survival$Cancer,levels = unique(BCAA.Survival$Cancer)) library(RColorBrewer) p1 <- ggplot()+ geom_segment(data = BCAA.Survival, aes(x=0,xend=log10(HR),y=Cancer,yend=Cancer))+ geom_point(data = BCAA.Survival, aes(x=log10(HR),y=Cancer,color=Color),size=3)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="log10(HR)",y="",color="")+ scale_color_manual(values = c(brewer.pal(8,"Set1")[1:2],"grey50"))+ scale_x_continuous(limits = c(-2,3))+ geom_vline(xintercept = c(0),linetype=2,color="black")+ coord_flip() p2 <- ggplot()+ geom_segment(data = BCAA.Survival, aes(x=0,xend=-log10(KM.pvalue), y=Cancer,yend=Cancer))+ geom_point(data = BCAA.Survival, aes(x=-log10(KM.pvalue), y=Cancer, color=-log10(KM.pvalue),size=-log10(KM.pvalue)))+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="-log10(P)",y="",color="")+ #scale_color_manual(values = c(brewer.pal(8,"Set1")[1:2],"grey50"))+ scale_color_gradientn(colors = brewer.pal(11,"RdBu")[4:1])+ scale_x_continuous(limits = c(0,13))+ geom_vline(xintercept = c(-log10(0.05)),linetype=2,color="black")+ coord_flip() ggsave(p2,filename = "BCAA.ssGSEA.Survival2.pdf",height = 2,width = 7) ggsave(p2,filename = "BCAA.ssGSEA.Survival2.Legend.pdf",height = 12,width = 7) ggsave(p,filename = "BCAA.ssGSEA.Survival.pdf",height = 2,width = 6) #ggsave(p,filename = "BCAA.Survival.Legend.pdf",height = 2,width = 16) ######## BCAA score Normal Tumor #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") BCAA.ID <- read.csv("BCAA.ID.csv") PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) %>% mutate(ProjectID=str_remove_all(Project.ID,"TCGA-")) %>% dplyr::select(ProjectID,SampleType,Sample.ID) PhenoData2 <- PhenoData %>% filter(SampleType=="Normal") BCAA.SCORE <- merge(BCAA.ssGSEA,PhenoData %>% filter(ProjectID %in% PhenoData2$ProjectID),by="Sample.ID") %>% data.frame(check.names = F) library(ggbeeswarm) library(RColorBrewer) library(ggpubr) data2 <- compare_means(formula = BCAA~SampleType,data = BCAA.SCORE,method = "wilcox.test", group.by="Cancer") %>% data.frame(check.names = F) write.csv(data2,file = "BCAA.ssGSEA.Tumor.Normal.Wilcox.csv",row.names = F) BCAA.SCORE2 <- BCAA.SCORE %>% group_by(Cancer,SampleType) %>% summarise(Mean=mean(BCAA)) %>% ungroup() %>% spread(SampleType,Mean,fill=NA) %>% mutate(SHAPE=if_else(Normal > Tumor,"Higher in normal","Higher in tumor")) %>% mutate(Normal=-Normal) %>% gather(Type,BCAA,Normal:Tumor) #%>% mutate(Cancer=factor(Cancer,levels = rev(unique(sort(.$Cancer))))) p1 <- ggplot(data=BCAA.SCORE2, aes(x=BCAA,y=Cancer))+ geom_rect(aes(xmin=0, xmax=Inf, ymin=-Inf, ymax=Inf),fill="#E64B35FF",alpha = 0.005)+ geom_rect(aes(xmin=-Inf, xmax=0, ymin=-Inf, ymax=Inf),fill="#00A087FF",alpha = 0.005)+ geom_segment(aes(x=0,xend=BCAA,y=Cancer,yend=Cancer, color=Type))+ geom_point(aes(fill=Type),color="black", size=3,shape=21)+ #geom_text(data=data2,aes(x=10.5,y=Cancer,label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="Mean of BCAA score",y="")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_shape_manual(values = c(21,22))+ scale_color_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ scale_fill_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ geom_vline(xintercept = c(0),color="black")+ scale_x_continuous(limits = c(-10,10)) #coord_cartesian(clip = "off")+ #geom_text(data = data2,aes(x=11,y=Cancer,label=paste("P=",round(p.format,3),sep = "")))> data2$label <- paste("P=",data2$p.format,sep = "") p2 <- ggplot(data2,aes(x=1,y=Cancer))+ geom_text(aes(label=p.signif))+ theme_void() p12 <- p1 %>% aplot::insert_right(p2,width = 0.25) ggsave(p12,filename = "BCAA.ssGSEA.Tumor.Normal2.pdf",height = 4,width = 3.5) ######## BCAA score Index Spearman #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(readxl) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") %>% mutate(ID=substr(Sample.ID,1,15)) #### ESTIMATE #### library(readxl) data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 2) data2<-data[,-1] Index.data2 <- merge(BCAA.ssGSEA,data2,by="ID") %>% unique() %>% data.frame() %>% gather(Attribute,Score,StromalScore:ESTIMATEScore) Spearman.Data <- data.frame() for (project in unique(Index.data2$Cancer)) { for (score in c("StromalScore","ImmuneScore","ESTIMATEScore")) { middata <- Index.data2 %>% filter(Attribute == score) %>% filter(Cancer==project) for (gene in c("BCAA")) { test <- cor.test(middata[[gene]],middata[["Score"]], method = "spearman", alternative = "two.sided", exact = F) Spearman.Data <- data.frame(Index=score, SCORE=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } } } write.csv(Spearman.Data,file = "BCAA.ESTIMATE.Spearman.csv",row.names = F) #### IPS #### library(readxl) data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 3) data2<-data[,-1] Index.data2 <- merge(BCAA.ssGSEA,data2,by="ID") %>% unique() %>% data.frame() %>% gather(Attribute,Score,MHC:IPS) for (project in unique(Index.data2$Cancer)) { for (score in c("AZ","CP","EC","MHC","SC","IPS")) { middata <- Index.data2 %>% filter(Attribute == score) %>% filter(Cancer==project) for (gene in c("BCAA")) { test <- cor.test(middata[[gene]],middata[["Score"]], method = "spearman", alternative = "two.sided", exact = F) Spearman.Data <- data.frame(Index=score, SCORE=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } } } write.csv(Spearman.Data,file = "BCAA.IPS.Spearman.csv",row.names = F) #### TMB MSI NEO HRD #### for (index in c(4,6,7,10)) { data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = index) data2<-data[,-1] Index.data2 <- merge(BCAA.ssGSEA,data2,by.x="ID",by.y="SampleName") %>% unique() %>% data.frame() for (project in unique(Index.data2$Cancer)) { middata <- Index.data2 %>% #filter(Attribute == score) %>% filter(Cancer==project) for (gene in c("BCAA")) { test <- cor.test(middata[[gene]],middata[[colnames(data2)[1]]], method = "spearman", alternative = "two.sided", exact = F) Spearman.Data <- data.frame(Index=colnames(data2)[1], SCORE=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } } } write.csv(Spearman.Data,file = "BCAA.TMB.MSI.NEO.HRD.Spearman.csv",row.names = F) #### MATH PURITY PLOIDY #### for (index in c(5,8,9)) { data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = index) data2<-data[,-1] Index.data2 <- merge(BCAA.ssGSEA,data2,by.x="ID",by.y="SampleName") %>% unique() %>% data.frame() for (project in unique(Index.data2$Cancer)) { middata <- Index.data2 %>% #filter(Attribute == score) %>% filter(Cancer==project) for (gene in c("BCAA")) { test <- cor.test(middata[[gene]],middata[[colnames(data2)[1]]], method = "spearman", alternative = "two.sided", exact = F) Spearman.Data <- data.frame(Index=colnames(data2)[1], SCORE=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } } } write.csv(Spearman.Data,file = "BCAA.MATH.PURITY.PLOIDY.Spearman.csv",row.names = F) #### RNA stem score #### for (index in c(13)) { data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = index) data2<-data[,-1] Index.data2 <- merge(BCAA.ssGSEA,data2,by.x="ID",by.y="SampleName") %>% unique() %>% data.frame() for (project in unique(Index.data2$Cancer)) { middata <- Index.data2 %>% #filter(Attribute == score) %>% filter(Cancer==project) for (gene in c("BCAA")) { test <- cor.test(middata[[gene]],middata[[colnames(data2)[1]]], method = "spearman", alternative = "two.sided", exact = F) Spearman.Data <- data.frame(Index=colnames(data2)[1], SCORE=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } } } write.csv(Spearman.Data,file = "BCAA.RNAss.Spearman.csv",row.names = F) ##### Figure #### Spearman.Data <- read.csv("BCAA.RNAss.Spearman.csv") write.csv(Spearman.Data,file = "BCAA.ssGSEA.Index.Spearman.csv",row.names = F) Spearman.Data$Index <- factor(Spearman.Data$Index,levels = unique(Spearman.Data$Index)) Spearman.Data$Cancer <- factor(Spearman.Data$Cancer,levels = sort(unique(Spearman.Data$Cancer))) library(RColorBrewer) Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[1:11], brewer.pal(8,"Dark2")[1:6]) p12 <- ggplot()+ geom_tile(data=Spearman.Data, aes(x=Cancer,y=Index,fill=Rho), color="white")+ geom_point(data=Spearman.Data %>% filter(Pvalue <= 0.05), aes(x=Cancer,y=Index,size=-log10(Pvalue)), shape=1,color="black")+ geom_point(data=Spearman.Data %>% filter(Pvalue > 0.05), aes(x=Cancer,y=Index,size=-log10(Pvalue)), shape=4,color="white")+ scale_size_continuous(range = c(1,4))+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="Spearman'CC")+ #geom_vline(xintercept = c(0),linetype=2)+ #theme(strip.text.x = element_text(size = 12))+ #ggthemes::scale_fill_gradient2_tableau(palette = "Red-Blue Diverging") scale_fill_gradient2(low = "#0066CC", mid = "white", high = "#c72e29") ggsave(p12,filename = "BCAA.ssGSEA.Index.Spearman2.pdf",height = 4,width = 8) ######## BCAA score ImmuneCellAI TIMER ... Spearman #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") BCAA.ID <- read.csv("BCAA.ID.csv") TIMER2 <- read.csv("../../TIMER2/infiltration_estimation_for_tcga.csv",check.names = F) PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) Immune.Data <- TIMER2 %>% #dplyr::select(c("cell_type",ends_with("CIBERSORT"))) %>% remove_rownames() %>% column_to_rownames("cell_type") %>% #magrittr::set_colnames(str_remove_all(colnames(.),"_CIBERSORT")) %>% rownames_to_column("SampleID") IMMUNECELL.SYMBOL.SPEARMAN <- data.frame() for (project in unique(PhenoType$Cancer)) { print(project) mid.Pheno <- PhenoType %>% filter(Cancer==project) %>% filter(Type=="Tumor") Exp.Filename <- paste("TCGA-",project,".mRNA.TPM.csv",sep = "") Exp.Data <- read.csv(file.path("../../mRNA.PanCancer.Exp/",Exp.Filename), row.names = 1,check.names = F) Exp.Data2 <- log2(Exp.Data+1) %>% data.frame(check.names = F) %>% rownames_to_column("ENSEMBL") %>% merge(BCAA.ID,by="ENSEMBL") %>% dplyr::select(-ENSEMBL,-ENTREZ) %>% column_to_rownames("SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% filter(Sample.ID %in% mid.Pheno$Sample.ID) %>% merge(BCAA.ssGSEA,by = "Sample.ID") %>% mutate(SampleID=substr(Sample.ID,1,15)) %>% merge(Immune.Data,by="SampleID") for (gene1 in c("BCAA")) { for (gene2 in colnames(Immune.Data)[2:ncol(Immune.Data)]) { if (sum(! is.na(Exp.Data2[[gene2]])) >= 5) { test <- cor.test(Exp.Data2[[gene1]],Exp.Data2[[gene2]], method = "spearman", exact = F, alternative = "two.sided") IMMUNECELL.SYMBOL.SPEARMAN <- data.frame(IMMUNECELL=gene2, SYMBOL=gene1, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(IMMUNECELL.SYMBOL.SPEARMAN) }else{ IMMUNECELL.SYMBOL.SPEARMAN <- data.frame(IMMUNECELL=gene2, SYMBOL=gene1, Rho=NA, Pvalue=NA, Cancer=project) %>% rbind.data.frame(IMMUNECELL.SYMBOL.SPEARMAN) } } } } write.csv(IMMUNECELL.SYMBOL.SPEARMAN,file = "BCAA.ssGSEA.IMMUNECELL.SPEARMAN.csv",row.names = F) #### Figure 1 #### IMMUNECELL.SYMBOL.SPEARMAN <- read.csv("BCAA.ssGSEA.IMMUNECELL.SPEARMAN.csv") IMMUNECELL.SYMBOL.SPEARMAN2 <- IMMUNECELL.SYMBOL.SPEARMAN %>% filter(str_detect(IMMUNECELL,"score")) %>% mutate(Method=str_remove_all(IMMUNECELL,".*_")) %>% mutate(ImmuneCell=str_remove_all(IMMUNECELL,"_.*")) %>% mutate(ImmuneCell=Hmisc::capitalize(ImmuneCell)) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% mutate(ImmuneCell=factor(ImmuneCell,levels = c("Microenvironment score", "Stroma score", "Immune score", "Cytotoxicity score"))) Figure.Body <- ggplot()+ geom_point(data=IMMUNECELL.SYMBOL.SPEARMAN2, aes(x=ImmuneCell,y=Cancer, size=-log10(Pvalue), fill=Rho), shape=21,color="white")+ geom_point(data=IMMUNECELL.SYMBOL.SPEARMAN2 %>% filter(Pvalue<=0.05), aes(x=ImmuneCell,y=Cancer, size=-log10(Pvalue), fill=Rho), shape=21,color="black",stroke=1.2)+ #facet_wrap(~SCORE)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="Spearman'CC")+ #geom_vline(xintercept = c(0),linetype=2)+ #theme(strip.text.x = element_text(size = 12))+ #ggthemes::scale_fill_gradient2_tableau(palette = "Red-Blue Diverging") scale_fill_gradient2(low = "#0066CC", mid = "white", high = "#c72e29") Figure.Top <- IMMUNECELL.SYMBOL.SPEARMAN2 %>% dplyr::select(ImmuneCell,Method) %>% unique() %>% mutate(ImmuneCell=factor(ImmuneCell,levels = c("Microenvironment score", "Stroma score", "Immune score", "Cytotoxicity score"))) %>% ggplot(aes(ImmuneCell,y=1,fill=Method))+ geom_col(width = 1,color="black")+ theme_void()+ scale_fill_manual(values = brewer.pal(8,"Set1")[2:1]) Figure.F <- Figure.Body %>% aplot::insert_top(Figure.Top,height = 0.03) ggsave(Figure.F,filename = "BCAA.ssGSEA.ImmuneScore.Spearman.pdf",height = 6,width = 3.5) #### Figure 2 #### IMMUNECELL.SYMBOL.SPEARMAN <- read.csv("BCAA.ssGSEA.IMMUNECELL.SPEARMAN.csv") IMMUNECELL.SYMBOL.SPEARMAN3 <- IMMUNECELL.SYMBOL.SPEARMAN %>% filter(!str_detect(IMMUNECELL,"score")) %>% filter(!str_detect(IMMUNECELL,"uncharacter")) %>% mutate(Method=str_remove_all(IMMUNECELL,".*_")) %>% filter(! str_detect(IMMUNECELL,"EPIC")) %>% filter(Method != "CIBERSORT-ABS") %>% mutate(ImmuneCell=str_remove_all(IMMUNECELL,"_.*")) %>% mutate(ImmuneCell=Hmisc::capitalize(ImmuneCell)) %>% mutate(Label=if_else(Pvalue > 0.05,"",if_else(Pvalue > 0.01,"*",if_else(Pvalue > 0.001,"**","***")))) %>% mutate(Cancer=factor(Cancer,levels = sort((unique(.$Cancer))))) %>% arrange(desc(ImmuneCell),Method) %>% mutate(IMMUNECELL=factor(IMMUNECELL,levels = unique(.$IMMUNECELL))) #mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) Figure.Body <- ggplot() + geom_tile(data=IMMUNECELL.SYMBOL.SPEARMAN3, aes(x=Cancer,y=IMMUNECELL,fill=Rho), color="white")+ geom_point(data=IMMUNECELL.SYMBOL.SPEARMAN3 %>% filter(Pvalue <= 0.05), aes(x=Cancer,y=IMMUNECELL,size=-log10(Pvalue)), shape=1,color="black")+ geom_point(data=IMMUNECELL.SYMBOL.SPEARMAN3 %>% filter(Pvalue > 0.05), aes(x=Cancer,y=IMMUNECELL,size=-log10(Pvalue)), shape=4,color="white")+ scale_size_continuous(range = c(1,3))+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="Spearman'CC")+ scale_fill_gradient2(low = "#0066CC", mid = "white", high = "#c72e29") #ggsave(Figure.Body,filename = "BCAA.ssGSEA.ImmuneCell.Spearman2.pdf",width = 5.5,height = 6) Figure.R1 <- IMMUNECELL.SYMBOL.SPEARMAN3 %>% dplyr::select(IMMUNECELL,Method) %>% unique() %>% mutate(IMMUNECELL=factor(IMMUNECELL,levels = unique(.$IMMUNECELL))) %>% ggplot(aes(1,y=IMMUNECELL,fill=Method))+ geom_col(width = 1,color="black")+ theme_void()+ #scale_fill_manual(values = brewer.pal(8,"Paired")[1:6]) ggsci::scale_fill_npg() Figure.R2 <- IMMUNECELL.SYMBOL.SPEARMAN3 %>% mutate(Correlation=if_else(Pvalue > 0.05,"p>0.05", if_else(Rho > 0,"Positive","Negative"))) %>% group_by(IMMUNECELL,Correlation) %>% dplyr::summarise(Count=n()) %>% #arrange(desc(Count)) ungroup() %>% #mutate(Yaxis=paste(miRNA,Target,sep = "->")) %>% #dplyr::select(-Target,-miRNA) %>% spread(Correlation,Count,fill=0) %>% gather(Correlation,Count,-IMMUNECELL) %>% mutate(Correlation=factor(Correlation, levels = c("p>0.05","Negative","Positive"))) %>% ggplot(aes(x=Count,y=IMMUNECELL,fill=Correlation))+ geom_col(width = 1,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_blank(), axis.ticks.y=element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="No. of cancers",y="")+ scale_fill_manual(values = c("grey90","#66C2A5","#BC3C2999")) Figure.F <- Figure.Body %>% aplot::insert_right(Figure.R1,width = 0.05) %>% aplot::insert_right(Figure.R2,width = 0.3) ggsave(Figure.F,filename = "BCAA.ssGSEA.ImmuneCell.Spearman.pdf",width = 10,height = 12) ######## BCAA score HALLMARKS #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") BCAA.ID <- read.csv("BCAA.ID.csv") TERM <- c("HALLMARK_E2F_TARGETS", "HALLMARK_G2M_CHECKPOINT", "HALLMARK_MYC_TARGETS_V2", "HALLMARK_MYC_TARGETS_V1", "HALLMARK_MITOTIC_SPINDLE", "HALLMARK_NOTCH_SIGNALING", "HALLMARK_WNT_BETA_CATENIN_SIGNALING", "HALLMARK_MTORC1_SIGNALING", "HALLMARK_TNFA_SIGNALING_VIA_NFKB", "HALLMARK_TGF_BETA_SIGNALING", "HALLMARK_KRAS_SIGNALING_UP", "HALLMARK_KRAS_SIGNALING_DN", "HALLMARK_IL6_JAK_STAT3_SIGNALING", "HALLMARK_INTERFERON_GAMMA_RESPONSE", "HALLMARK_INTERFERON_ALPHA_RESPONSE", "HALLMARK_INFLAMMATORY_RESPONSE", "HALLMARK_HEDGEHOG_SIGNALING", "HALLMARK_COMPLEMENT", "HALLMARK_DNA_REPAIR", "HALLMARK_HYPOXIA", "HALLMARK_APOPTOSIS", "HALLMARK_P53_PATHWAY", "HALLMARK_PI3K_AKT_MTOR_SIGNALING", "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION") Spearman.HALLMARK <- data.frame() for (project in unique(BCAA.ssGSEA$Cancer)) { mid.BCAA <- BCAA.ssGSEA %>% filter(Cancer==project) filename <- paste("TCGA-",project,".ssGSEA.csv",sep = "") ssGSEA.Exp <- read.csv(file.path("../../HALLMARK.ssGSEA.Pancancer/",filename),check.names = F,row.names = 1) Spearman.Data <- ssGSEA.Exp %>% rownames_to_column("TERM") %>% filter(TERM %in% TERM) %>% remove_rownames() %>% column_to_rownames("TERM") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% merge(mid.BCAA,by="Sample.ID") #Spearman.HALLMARK <- rbind.data.frame(Spearman.Data,Spearman.HALLMARK) for (HALLMARK in TERM) { for (GENE in c("BCAA")) { test <- cor.test(Spearman.Data[[HALLMARK]], Spearman.Data[[GENE]], method = "spearman", alternative = "two.sided", exact = F) Spearman.HALLMARK <- data.frame(HALLMARK = HALLMARK, SCORE = GENE, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.HALLMARK) } } } write.csv(Spearman.HALLMARK,"BCAA.ssGSEA.HALLMARK.Spearman.csv",row.names = F) #### Figure #### Spearman.HALLMARK <- read.csv("BCAA.ssGSEA.HALLMARK.Spearman.csv") Spearman.HALLMARK$HALLMARK <- str_remove_all(Spearman.HALLMARK$HALLMARK,"HALLMARK_") library(tidyverse) library(RColorBrewer) Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[1:11], brewer.pal(8,"Dark2")[1:6]) names(Cancer.Col) <- sort(unique(Spearman.HALLMARK$Cancer)) p1 <- ggplot(Spearman.HALLMARK, aes(x=Rho,y=reorder(HALLMARK,Rho), size=-log10(Pvalue), color=Cancer))+ geom_point()+ #facet_wrap(~SCORE)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=9), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="Spearman'CC",y="")+ geom_vline(xintercept = c(0),linetype=2)+ theme(strip.text.x = element_text(size = 12))+ scale_color_manual(values = Cancer.Col) Spearman.HALLMARK2 <- Spearman.HALLMARK %>% #filter(Pvalue<=0.05) %>% mutate(Interaction=if_else(Rho>0,"Positive","Negative")) %>% group_by(HALLMARK,Interaction,SCORE) %>% summarise(Sig.Count=n()) %>% ungroup() %>% spread(Interaction,Sig.Count,fill=0) %>% magrittr::set_colnames(c("HALLMARK","SCORE","S.N","S.P")) Spearman.HALLMARK3 <- Spearman.HALLMARK %>% filter(Pvalue<=0.05) %>% mutate(Interaction=if_else(Rho>0,"Positive","Negative")) %>% group_by(HALLMARK,Interaction,SCORE) %>% summarise(Sig.Count=n()) %>% ungroup() %>% spread(Interaction,Sig.Count,fill=0) %>% magrittr::set_colnames(c("HALLMARK","SCORE","Sig.N","Sig.P")) %>% merge(Spearman.HALLMARK2,by=c("HALLMARK","SCORE")) %>% mutate(E.P=S.P-Sig.P) %>% mutate(E.N=S.N-Sig.N) %>% arrange(S.P) %>% mutate(Sig.N=-Sig.N,E.N=-E.N) %>% #mutate(HALLMARK=factor(HALLMARK,levels = .$HALLMARK)) %>% dplyr::select(-S.P,-S.N) %>% gather(Attribute,Count,Sig.N:E.N) p2 <- ggplot(Spearman.HALLMARK3,aes(x=Count,y=reorder(HALLMARK,Count),fill=Attribute))+ geom_col()+ ggthemes::theme_few()+ facet_wrap(~SCORE)+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(y="",x="No. of cancer types",fill="Type")+ #scale_fill_discrete()+ scale_fill_manual(breaks=c("E.N","E.P","Sig.N","Sig.P"), labels=c("Insignificant negative correlation", "Insignificant positive correlation", "Significant negative correlation", "Significant positive correlation"), values = c("#4DAF4A","#FF7F00","#377EB8","#E41A1C")) Figure.F <- p1 %>% aplot::insert_right(p2,width = 1) ggsave(Figure.F,filename = "BCAA.ssGSEA.HALLMARK.Spearman.pdf",height = 5,width = 8) ggsave(Figure.F,filename = "BCAA.ssGSEA.HALLMARK.Spearman.Legend.pdf",height = 10,width = 8) ######## BCAA score PATHWAY Spearman ############ setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) %>% mutate(ProjectID=str_remove_all(Project.ID,"TCGA-")) %>% dplyr::select(ProjectID,SampleType,Sample.ID) %>% filter(SampleType=="Tumor") BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") %>% mutate(barcode = substring(Sample.ID,1,12)) %>% filter(Sample.ID %in% PhenoData$Sample.ID) BCAA.ID <- read.csv("BCAA.ID.csv") Cancer.Score.PAS.Spearman <- data.frame() files <- list.files(path = "../../GSCA/",pattern = "tsv.gz") files <- files[str_detect(files,"PAS")] for (file in files) { project = file %>% str_remove_all("\\..*") library(data.table) data <- fread(file.path("../../GSCA/",file),sep = "\t",check.names = F) Exp.PAS <- merge(data,BCAA.ssGSEA,by="barcode") for (index in unique(Exp.PAS$pathway)) { mid.data <- Exp.PAS %>% filter(pathway==index) for (gene in c("BCAA")) { test <- cor.test(mid.data[[gene]],mid.data[["score"]], method = "spearman", alternative = "two.sided", exact = F) Cancer.Score.PAS.Spearman <- data.frame(Cancer=project, SYMBOL=gene, PATHWAY=index, Rho=test$estimate, Pvalue=test$p.value, Count=nrow(mid.data)) %>% rbind.data.frame(Cancer.Score.PAS.Spearman) } } } write.csv(Cancer.Score.PAS.Spearman,file = "BCAA.ssGSEA.PAS.Spearman.csv",row.names = F) #### Figure #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) Cancer.Score.PAS.Spearman <- read.csv(file = "BCAA.ssGSEA.PAS.Spearman.csv") Cancer.Score.PAS.Spearman2 <- Cancer.Score.PAS.Spearman %>% mutate(PATHWAY=factor(PATHWAY,levels = rev(sort(unique(.$PATHWAY))))) %>% arrange(Cancer,SYMBOL) %>% rstatix::add_significance("Pvalue") %>% mutate(label=if_else(Pvalue.signif == "ns","",Pvalue.signif)) %>% mutate(Xaxis=paste(Cancer,"(",Count,")",sep = "")) p21 <- ggplot()+ geom_tile(data=Cancer.Score.PAS.Spearman2, aes(x=Xaxis,y=PATHWAY,fill=Rho), color="white")+ geom_point(data=Cancer.Score.PAS.Spearman2 %>% filter(Pvalue <= 0.05), aes(x=Xaxis,y=PATHWAY,size=-log10(Pvalue)), shape=1,color="black")+ geom_point(data=Cancer.Score.PAS.Spearman2 %>% filter(Pvalue > 0.05), aes(x=Xaxis,y=PATHWAY,size=-log10(Pvalue)), shape=4,color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradient2("Spearman'CC", low = "#053061",high="#67001F",mid = "white") ggsave(p21,filename = "BCAA.ssGSEA.PAS.Spearman.pdf",height = 4,width = 8) ######## BCAA score Subtype #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") BCAA.ID <- read.csv("BCAA.ID.csv") PhenoData <- read.csv("../../mRNA.Phenotype.csv") %>% filter(Sample.Type %in% c("Primary Tumor", "Primary Blood Derived Cancer - Peripheral Blood", "Solid Tissue Normal")) %>% mutate(SampleType = ifelse(Sample.Type == "Solid Tissue Normal","Normal","Tumor")) %>% mutate(ProjectID=str_remove_all(Project.ID,"TCGA-")) %>% dplyr::select(ProjectID,SampleType,Sample.ID) %>% filter(SampleType=="Tumor") BCAA.SCORE <- merge(BCAA.ssGSEA,PhenoData,by="Sample.ID") %>% data.frame(check.names = F) %>% mutate(barcode=substr(Sample.ID,1,12)) files <- list.files(path = "../../GSCA/",pattern = "gz$") files <- files[str_detect(files,"subtype")] Subtype.Data <- data.frame() library(data.table) for (file in files) { project <- file %>% str_remove_all("\\..*") Subtype.Data <- fread(file.path("../../GSCA/",file)) %>% merge(BCAA.SCORE,by="barcode") %>% mutate(Cancer=project) %>% rbind.data.frame(Subtype.Data) } write.csv(Subtype.Data,file = "BCAA.ssGSEA.SubType.Data.csv",row.names = F) data2 <- compare_means(formula = BCAA~subtype,data = Subtype.Data,method = "kruskal.test", group.by="Cancer") %>% data.frame(check.names = F) write.csv(data2,file = "BCAA.ssGSEA.SubType.Data.Kruskal.csv",row.names = F) #### Figure #### Subtype.Data <- read.csv("BCAA.ssGSEA.SubType.Data.csv") data2 <- read.csv("BCAA.ssGSEA.SubType.Data.Kruskal.csv") Subtype.Data2 <- Subtype.Data %>% arrange(desc(Cancer),subtype) %>% mutate(Yaxis=paste(Cancer,subtype,sep = ":")) %>% mutate(Yaxis=factor(Yaxis,levels = unique(Yaxis))) %>% group_by(Yaxis) %>% summarise(Mean=mean(BCAA)) %>% mutate(Cancer=str_remove_all(Yaxis,":.*")) Subtype.Data3 <- Subtype.Data %>% arrange(desc(Cancer),subtype) %>% mutate(Yaxis=paste(Cancer,subtype,sep = ":")) %>% mutate(Yaxis=factor(Yaxis,levels = unique(Yaxis))) p10 <- ggplot(Subtype.Data3,aes(x=BCAA,y=Yaxis,fill=Cancer))+ stat_summary(fun.data = "mean_cl_boot", # # 误差线 geom = "errorbar", width = .4) + stat_summary(fun = "mean", geom = "point",shape=21)+ labs(x="",y="")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=12), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_fill_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_x_continuous(trans = "log10")+ #ggsci::scale_fill_npg() ggthemes::scale_fill_stata()+ scale_x_continuous(limits = c(0,10),breaks = c(0,3,6,9)) p10.F <- p10 %>% aplot::insert_right(p10.R,width = 0.15) ggsave(p10.F,filename = "BCAA.ssGSEA.Subtype.Tumor2.pdf",height = 5,width = 4) p10<- ggplot(data=Subtype.Data2, aes(x=Mean,y=Yaxis,fill=Cancer))+ geom_segment(aes(x=0,xend=Mean,y=Yaxis,yend=Yaxis),color="black")+ geom_point(size=3,shape=21,color="black")+ #geom_boxplot()+ #geom_text(data=data2,aes(x=10.5,y=Cancer,label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), axis.text.y = element_text(size=12), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_fill_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_x_continuous(trans = "log10")+ #ggsci::scale_fill_npg() ggthemes::scale_fill_stata()+ scale_x_continuous(limits = c(0,10),breaks = c(0,3,6,9)) p10.R <- Subtype.Data2 %>% dplyr::select(Yaxis,Cancer) %>% unique() %>% mutate(Yaxis=factor(Yaxis,levels = unique(Yaxis))) %>% ggplot(aes(x=1,y=Yaxis,fill=Cancer))+ geom_col(width=1)+ theme_void()+ #ggsci::scale_fill_nejm() ggthemes::scale_fill_stata() p10.F <- p10 %>% aplot::insert_right(p10.R,width = 0.15) ggsave(p10.F,filename = "BCAA.ssGSEA.Subtype.Tumor.pdf",height = 5,width = 4) ######## BCAA score Grade #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") %>% mutate(SampleName=substring(Sample.ID,1,15)) BCAA.ID <- read.csv("BCAA.ID.csv") library(readxl) data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 18) data2<-data[,-1] BCAA.GRADE <- merge(BCAA.ssGSEA,data2,by="SampleName") data2 <- compare_means(formula = BCAA~Grade,data = BCAA.GRADE,method = "kruskal.test", group.by="Cancer") %>% data.frame(check.names = F) write.csv(BCAA.GRADE,file = "BCAA.ssGSEA.Grade.Data.csv",row.names = F) write.csv(data2,file = "BCAA.ssGSEA.Grade.Data.Kruskal.csv",row.names = F) BCAA.GRADE <- read.csv("BCAA.ssGSEA.Grade.Data.csv") data2 <- read.csv("BCAA.ssGSEA.Grade.Data.Kruskal.csv") BCAA.GRADE2 <- BCAA.GRADE %>% mutate(Xaxis=paste(Cancer,Grade,sep = "_")) p11<- ggplot(data=BCAA.GRADE2, aes(x=Xaxis,y=BCAA, #Cancer,# fill=Cancer))+ #geom_boxplot()+ #geom_text(data=data2,aes(x=10.5,y=Cancer,label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ stat_summary(fun.data = "mean_cl_boot", # # 误差线 geom = "errorbar", width = .4) + stat_summary(fun = "mean", geom = "point",shape=21)+ labs(x="",y="BCAA score")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust=0.5), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_fill_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_x_continuous(trans = "log10")+ #ggsci::scale_fill_npg() ggthemes::scale_fill_stata() p11.T <- BCAA.GRADE2 %>% dplyr::select(Xaxis,Cancer) %>% unique() %>% #mutate(Xaxis=factor(Xaxis,levels = unique(Xaxis))) %>% ggplot(aes(x=Xaxis,y=1,fill=Cancer))+ geom_col(width=1)+ theme_void()+ #ggsci::scale_fill_nejm() ggthemes::scale_fill_stata()+ theme(legend.position = "bottom") p11.F <- p11 %>% aplot::insert_top(p11.T,height = 0.15) ggsave(p11.F,filename = "BCAA.ssGSEA.Grade.Tumor2.pdf",height = 2.5,width = 8) ggsave(p11.F,filename = "BCAA.ssGSEA.Grade.Tumor2.Ledend.pdf",height = 12,width = 8) ######## BCAA score Stage #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") %>% mutate(SampleName=substring(Sample.ID,1,15)) BCAA.ID <- read.csv("BCAA.ID.csv") library(readxl) data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 19) data2<-data[,-1] BCAA.STAGE <- merge(BCAA.ssGSEA,data2,by="SampleName") data2 <- compare_means(formula = BCAA~Stage, data = BCAA.STAGE, method = "kruskal.test", group.by="Cancer") %>% data.frame(check.names = F) write.csv(BCAA.STAGE,file = "BCAA.ssGSEA.BCAA.STAGE.Data.csv",row.names = F) write.csv(data2,file = "BCAA.ssGSEA.BCAA.STAGE.Data.Kruskal.csv",row.names = F) BCAA.STAGE <- read.csv("BCAA.ssGSEA.BCAA.STAGE.Data.csv") data2 <- read.csv("BCAA.ssGSEA.BCAA.STAGE.Data.Kruskal.csv") BCAA.STAGE2 <- BCAA.STAGE %>% mutate(STAGE=str_remove_all(Stage,"Stage ")) %>% mutate(Xaxis=paste(Cancer,STAGE,sep = "_")) p11<- ggplot(data=BCAA.STAGE2, aes(x=Xaxis,y=BCAA, #Cancer,# fill=Cancer))+ #geom_boxplot()+ #geom_text(data=data2,aes(x=10.5,y=Cancer,label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ stat_summary(fun.data = "mean_cl_boot", # # 误差线 geom = "errorbar", width = .4) + stat_summary(fun = "mean", geom = "point",shape=21)+ labs(x="",y="BCAA score")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_blank(),#element_text(size=9,angle = 90,vjust=0.5), axis.text.y = element_text(size=9), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_fill_manual("",values = brewer.pal(9,"Set1")[c(2:1)])+ #scale_x_continuous(trans = "log10")+ #ggsci::scale_fill_npg() #ggthemes::scale_fill_stata() scale_fill_manual(values = Cancer.Col) p11.T1 <- BCAA.STAGE2 %>% dplyr::select(Xaxis,Cancer) %>% unique() %>% #mutate(Xaxis=factor(Xaxis,levels = unique(Xaxis))) %>% ggplot(aes(x=Xaxis,y=1,fill=Cancer))+ geom_col(width=1)+ theme_void()+ #ggsci::scale_fill_nejm() #ggthemes::scale_fill_stata()+ scale_fill_manual(values = Cancer.Col)+ theme(legend.position = "bottom") p11.T2 <- BCAA.STAGE2 %>% dplyr::select(Xaxis,STAGE) %>% unique() %>% #mutate(Xaxis=factor(Xaxis,levels = unique(Xaxis))) %>% ggplot(aes(x=Xaxis,y=1,fill=STAGE))+ geom_col(width=1)+ theme_void()+ ggsci::scale_fill_npg()+ #ggthemes::scale_fill_stata()+ #scale_fill_manual(values = Cancer.Col)+ theme(legend.position = "bottom") p11.F <- p11 %>% aplot::insert_top(p11.T2,height = 0.1) %>% aplot::insert_top(p11.T1,height = 0.1) ggsave(p11.F,filename = "BCAA.ssGSEA.STAGE.Tumor2.pdf",height = 2,width = 12) ggsave(p11.F,filename = "BCAA.ssGSEA.STAGE.Tumor2.Ledend.pdf",height = 12,width = 12) ######## DEGS => GSEA ######## #### DEGs #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) BCAA.ssGSEA <- read.csv("BCAA.ssGSEA/BCAA.ssGSEA.csv") %>% mutate(SampleName=substring(Sample.ID,1,15)) #BCAA.ssGSEA <- read.csv("BCAA.ssGSEA.csv") %>% # mutate(SampleName=substring(Sample.ID,1,15)) BCAA.ID <- read.csv("BCAA.ID.csv") library(GSVA) library(GSEABase) library(msigdbr) library(clusterProfiler) library(org.Hs.eg.db) library(enrichplot) library(limma) library(tidyverse) library(TCGAbiolinks) library(DESeq2) PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) Hallmark <- read.gmt("../../msigdb/h.all.v7.5.1.symbols.gmt") dir.create("DEGs.HALLMARK") files <- list.files("../../mRNA.Pancancer.Exp.Count/",pattern = "csv$") #files <- list.files("../mRNA.Pancancer.Exp.Count/",pattern = "csv$") for (file in files) { project <- file %>%str_remove("PanCancer\\.") %>% str_remove("\\..*") %>% str_remove("TCGA-") print(project) #Count.data <- read.csv(file = file.path("../../mRNA.Pancancer.Exp.Count/",file),check.names = F,row.names = 1) Count.data <- read.csv(file = file.path("../mRNA.Pancancer.Exp.Count/",file),check.names = F,row.names = 1) filename <- paste("TCGA-",project,".mRNA.TPM.csv",sep = "") TPM.data <- read.csv(file = file.path("../mRNA.PanCancer.Exp/",filename),check.names = F,row.names = 1) #TPM.data <- read.csv(file = file.path("../../mRNA.PanCancer.Exp/",filename),check.names = F,row.names = 1) mid.Pheno <- PhenoType %>% filter(Type=="Tumor") %>% filter(Cancer==project) for (gene in BCAA.ID$ENSEMBL) { print(gene) SYMBOL=(BCAA.ID%>%filter(ENSEMBL==gene))$SYMBOL outfilename <- paste(project,SYMBOL,"DEGs.csv",sep = ".") if (!file.exists(file.path("DEGs.HALLMARK",outfilename))) { data2 <- TPM.data %>% dplyr::select(intersect(colnames(.),mid.Pheno$Sample.ID)) GENE.EXP = data2[gene,] %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") GENE.EXP$Group=if_else(GENE.EXP[[gene]]>median(GENE.EXP[[gene]]),"high","low") DEG.data <- Count.data %>% rownames_to_column("ENSEMBL") %>% filter(ENSEMBL != gene) %>% column_to_rownames("ENSEMBL") %>% dplyr::select(GENE.EXP$Sample.ID) #High.Count <- DEG.data[,(GENE.EXP %>% filter(Group=="high"))$Sample.ID] #Low.Count <- DEG.data[,(GENE.EXP %>% filter(Group=="low"))$Sample.ID] countData <- DEG.data[rowMeans(DEG.data)>1,] condition <- factor(GENE.EXP$Group,levels = c("high","low")) colData <- data.frame(row.names=colnames(countData), condition) dds <- DESeqDataSetFromMatrix(countData = countData, colData = colData, design = ~ condition) dds1 <- DESeq(dds) res <- results(dds1,contrast=c("condition","high","low")) res1 <- data.frame(res, stringsAsFactors = FALSE, check.names = FALSE) write.csv(res1,file = file.path("DEGs.HALLMARK/",outfilename)) } } mid.BCAA <- BCAA.ssGSEA %>% filter(Cancer==project) %>% filter(Sample.ID %in% mid.Pheno$Sample.ID) for (gene in c("BCAA")) { print(gene) outfilename <- paste(project,gene,"DEGs.csv",sep = ".") if (!file.exists(file.path("DEGs.HALLMARK",outfilename))) { mid.BCAA$Group=if_else(mid.BCAA[[gene]]>median(mid.BCAA[[gene]]),"high","low") DEG.data <- Count.data[,mid.BCAA$Sample.ID] %>% data.frame(check.names = F) countData <- DEG.data[rowMeans(DEG.data)>1,] condition <- factor(mid.BCAA$Group,levels = c("high","low")) colData <- data.frame(row.names=colnames(countData), condition) dds <- DESeqDataSetFromMatrix(countData = countData, colData = colData, design = ~ condition) dds1 <- DESeq(dds) res <- results(dds1,contrast=c("condition","high","low")) res1 <- data.frame(res, stringsAsFactors = FALSE, check.names = FALSE) #SYMBOL=(BCAA.ID%>%filter(ENSEMBL==gene))$SYMBOL write.csv(res1,file = file.path("DEGs.HALLMARK/",outfilename)) } } } #### GSEA => HALLMARK #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(GSVA) library(GSEABase) library(msigdbr) library(fgsea) library(clusterProfiler) library(org.Hs.eg.db) library(enrichplot) library(limma) library(tidyverse) library(TCGAbiolinks) library(DESeq2) Hallmark <- read.gmt("../../msigdb/h.all.v7.5.1.symbols.gmt") GeneType <- read.table("../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,Gene.name) %>% unique() %>% na.omit() %>% filter(Gene.name != "") HALLMARK.GSEA <- data.frame() files <- list.files("DEGs.HALLMARK/",pattern = "csv$") for (file in files) { STRING <- file %>% str_split("\\.") %>% unlist() project = STRING[1] SYMBOL = STRING[2] print(project) data <- read.csv(file.path("DEGs.HALLMARK/",file),check.names = F,row.names = 1) data2 <- data %>% rownames_to_column("ENSEMBL") %>% merge(GeneType.Unique,by.x="ENSEMBL",by.y="Gene.stable.ID") %>% dplyr::select(-ENSEMBL) ge = data2$log2FoldChange names(ge) = data2$Gene.name ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =1) filename <- paste(project,SYMBOL,"GSEA.HALLMARK.Rdata",sep = ".") save(GSEA.Res,file = file.path("DEGs.HALLMARK/",filename)) HALLMARK.GSEA <- GSEA.Res@result %>% mutate(Cancer=project,SYMBOL=SYMBOL) %>% rbind.data.frame(HALLMARK.GSEA) } #View(HALLMARK.GSEA) write.csv(HALLMARK.GSEA,file = "DEGs.HALLMARK/DEGs.HALLMARK.GSEA.csv",row.names = F) #### GSEA => HALLMARK => Figure #### setwd("K:/TCGA/Anlysis/BCAA") library(ggDoubleHeat) library(ggplot2) library(tidyverse) library(readxl) BCAA.ID <- read.csv("BCAA.ID.csv") TERM <- c("HALLMARK_E2F_TARGETS", "HALLMARK_G2M_CHECKPOINT", "HALLMARK_MYC_TARGETS_V2", "HALLMARK_MYC_TARGETS_V1", "HALLMARK_MITOTIC_SPINDLE", "HALLMARK_NOTCH_SIGNALING", "HALLMARK_WNT_BETA_CATENIN_SIGNALING", "HALLMARK_MTORC1_SIGNALING", "HALLMARK_TNFA_SIGNALING_VIA_NFKB", "HALLMARK_TGF_BETA_SIGNALING", "HALLMARK_KRAS_SIGNALING_UP", "HALLMARK_KRAS_SIGNALING_DN", "HALLMARK_IL6_JAK_STAT3_SIGNALING", "HALLMARK_INTERFERON_GAMMA_RESPONSE", "HALLMARK_INTERFERON_ALPHA_RESPONSE", "HALLMARK_INFLAMMATORY_RESPONSE", "HALLMARK_HEDGEHOG_SIGNALING", "HALLMARK_COMPLEMENT", "HALLMARK_DNA_REPAIR", "HALLMARK_HYPOXIA", "HALLMARK_APOPTOSIS", "HALLMARK_P53_PATHWAY", "HALLMARK_PI3K_AKT_MTOR_SIGNALING", "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION") HALLMARK.GSEA <- read.csv("DEGs.HALLMARK/DEGs.HALLMARK.GSEA.csv") HALLMARK.GSEA.Data <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% mutate(Type=if_else(NES > 0,"High","Low")) %>% group_by(SYMBOL,ID,Type) %>% summarise(Count=n()) %>% ungroup() %>% spread(Type,Count,fill=0) %>% filter(ID %in% TERM) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) #%>% BCAA.ID <- read.csv("BCAA.ID.csv") HALLMARK.GSEA <- read.csv("DEGs.HALLMARK/DEGs.HALLMARK.GSEA.csv") HALLMARK.GSEA.Data <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% mutate(Type=if_else(NES > 0,"High","Low")) %>% group_by(SYMBOL,ID,Type) %>% summarise(Count=n()) %>% ungroup() %>% spread(Type,Count,fill=0) %>% #filter(ID %in% TERM) %>% #filter(str_detect(ID,"KEGG_") | str_detect(ID,"REACTOME_")) %>% #mutate(ID=str_remove_all(ID,"KEGG_")) filter(High >= 10 | Low >= 10) High.Cancer <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% filter(NES > 0) %>% dplyr::select(Cancer,SYMBOL,ID) %>% group_by(SYMBOL,ID) %>% mutate(High.Cancer=paste0(Cancer,collapse = ",")) %>% dplyr::select(-Cancer) %>% unique() Low.Cancer <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% filter(NES < 0) %>% dplyr::select(Cancer,SYMBOL,ID) %>% group_by(SYMBOL,ID) %>% mutate(Low.Cancer=paste0(Cancer,collapse = ",")) %>% dplyr::select(-Cancer) %>% unique() HALLMARK.GSEA.Data2 <- HALLMARK.GSEA.Data %>% merge(High.Cancer,by=c("ID","SYMBOL")) %>% merge(Low.Cancer,by=c("ID","SYMBOL")) %>% arrange(SYMBOL) write.csv(HALLMARK.GSEA.Data2,"TableS6.csv",row.names = F) #mutate(ID=factor(ID,levels = rev(TERM %>% str_remove_all("HALLMARK_")))) #mutate(SYMBOL = factor(SYMBOL,levels = c("BCAA","GATOR1","GATOR2",BCAA.ID$SYMBOL))) ##是这个R包的小hug，如果不改顺序后面颜色会乱 , 将用于作图的数据集，按照用于绘图的X轴数据列的首字母从小到大进行序列重排 p12 <- ggplot(data = HALLMARK.GSEA.Data, aes(x = SYMBOL, y = ID)) + geom_heat_tri(lower = High, upper = Low, lower_colors = RColorBrewer::brewer.pal(11,"RdBu")[6:1], upper_colors = RColorBrewer::brewer.pal(11,"RdBu")[6:11])+ ggthemes::theme_few()+ theme(legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.text.y = element_text(size=9), axis.title = element_blank(), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ geom_text(aes(label = Low), nudge_x = 0.22, nudge_y = 0.2,size= 3,color = c('#FF7F00'))+ geom_text(aes(label = High), nudge_x = -0.22, nudge_y = -0.2,size= 3,color = c('#1B9E77')) #scale_fill_gradientn(colours = RColorBrewer::brewer.pal(11,"RdBu")[6:11]) ggsave(p12,filename = "DEGs.HALLMARK.GSEA.pdf",height = 6,width = 8) ggsave(p12,filename = "DEGs.HALLMARK/DEGs.HALLMARK.GSEA.pdf",height = 6,width = 8) #### GSEA => KEGG #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(GSVA) library(GSEABase) library(msigdbr) library(fgsea) library(clusterProfiler) library(org.Hs.eg.db) library(enrichplot) library(limma) library(tidyverse) library(TCGAbiolinks) library(DESeq2) Hallmark <- read.gmt("../../msigdb/c2.all.v7.5.1.symbols.gmt") GeneType <- read.table("../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,Gene.name) %>% unique() %>% na.omit() %>% filter(Gene.name != "") HALLMARK.GSEA <- data.frame() files <- list.files("DEGs.HALLMARK/",pattern = "csv$") files <- files[str_detect(files,".DEGs.csv")] for (file in files) { STRING <- file %>% str_split("\\.") %>% unlist() project = STRING[1] SYMBOL = STRING[2] print(project) data <- read.csv(file.path("DEGs.HALLMARK/",file),check.names = F,row.names = 1) data2 <- data %>% rownames_to_column("ENSEMBL") %>% merge(GeneType.Unique,by.x="ENSEMBL",by.y="Gene.stable.ID") %>% dplyr::select(-ENSEMBL) ge = data2$log2FoldChange names(ge) = data2$Gene.name ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =1) filename <- paste(project,SYMBOL,"GSEA.KEGG.Rdata",sep = ".") save(GSEA.Res,file = file.path("DEGs.HALLMARK/",filename)) HALLMARK.GSEA <- GSEA.Res@result %>% mutate(Cancer=project,SYMBOL=SYMBOL) %>% rbind.data.frame(HALLMARK.GSEA) } #View(HALLMARK.GSEA) write.csv(HALLMARK.GSEA,file = "DEGs.HALLMARK/DEGs.KEGG.GSEA.csv",row.names = F) #### GSEA => KEGG => Figure #### setwd("K:/TCGA/Anlysis/BCAA") library(ggDoubleHeat) library(ggplot2) library(tidyverse) library(readxl) BCAA.ID <- read.csv("BCAA.ID.csv") HALLMARK.GSEA <- read.csv("DEGs.HALLMARK/DEGs.KEGG.GSEA.csv") HALLMARK.GSEA.Data <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% mutate(Type=if_else(NES > 0,"High","Low")) %>% group_by(SYMBOL,ID,Type) %>% summarise(Count=n()) %>% ungroup() %>% spread(Type,Count,fill=0) %>% #filter(ID %in% TERM) %>% filter(str_detect(ID,"KEGG_") | str_detect(ID,"REACTOME_")) %>% #mutate(ID=str_remove_all(ID,"KEGG_")) filter(High >= 10 | Low >= 10) High.Cancer <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% filter(NES > 0) %>% dplyr::select(Cancer,SYMBOL,ID) %>% group_by(SYMBOL,ID) %>% mutate(High.Cancer=paste0(Cancer,collapse = ",")) %>% dplyr::select(-Cancer) %>% unique() Low.Cancer <- HALLMARK.GSEA %>% filter(p.adjust <= 0.05) %>% filter(NES < 0) %>% dplyr::select(Cancer,SYMBOL,ID) %>% group_by(SYMBOL,ID) %>% mutate(Low.Cancer=paste0(Cancer,collapse = ",")) %>% dplyr::select(-Cancer) %>% unique() HALLMARK.GSEA.Data2 <- HALLMARK.GSEA.Data %>% merge(High.Cancer,by=c("ID","SYMBOL")) %>% merge(Low.Cancer,by=c("ID","SYMBOL")) %>% arrange(SYMBOL) write.csv(HALLMARK.GSEA.Data2,"TableS7.csv",row.names = F) ######## Risk score LASSO COX => glmnet => LASSO model #### library(glmnet) setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(survival) library(survminer) #BCAA.ssGSEA <- read.csv("BCAA.ssGSEA.csv") %>% # mutate(SampleName=substring(Sample.ID,1,15)) BCAA.ID <- read.csv("BCAA.ID.csv") files <- list.files(path = "../../mRNA.PanCancer.Exp/",pattern = "csv$") LASSO.Feature <- data.frame() for (file in files) { project <- file %>% str_remove_all("\\..*") %>% str_remove_all("TCGA-") print(project) phenotype <- read.csv(paste("../../Clinical.XENA/",project,"/","TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) #### survival datasets survival.data <- read.csv(paste("../../Clinical.XENA/",project,"/TCGA-",project,".survival.tsv",sep = ""),sep = "\t",header = T,check.names = F) Inter.Samples <- intersect(intersect(phenotype$submitter_id.samples,colnames(TPM)),survival.data$sample) #### symbol expression TPM.ENSEMBL.Target <- TPM.log[BCAA.ID$ENSEMBL,Inter.Samples] %>% t() %>% data.frame(check.names = F) %>% mutate(SampleID = rownames(.)) %>% merge(.,survival.data,by.x="SampleID",by.y="sample") x <- as.matrix(TPM.ENSEMBL.Target[,BCAA.ID$ENSEMBL]) #gene => column y <- Surv(as.double(TPM.ENSEMBL.Target$OS.time),as.double(TPM.ENSEMBL.Target$OS)) Select.Variables <- c() for (i in 1:500) { print(i) #fit <- glmnet(x, y, family="cox") #plot(fit,xvar = "lambda") #using 10-fold CV to select lambda: cv.fit =cv.glmnet(x, y, family="cox", nfolds=5) #cv.fit$lambda.min fit_regL=cv.fit fit_regL_CVdev=cv.fit # final Lasso model: model_lasso_min <- glmnet(x=x, y=y,family="cox",lambda=cv.fit$lambda.min) fit_regL_coef <- coef(fit_regL, s=fit_regL_CVdev$lambda.min) ### Plot Cross-Validation LASSO model #par(mfrow=c(1,2)) #plot(fit_regL_CVdev,las=1, main="Lasso fit CV") #abline(v=log(fit_regL_CVdev$lambda.min), col="orange", lwd=2) #text(log(fit_regL_CVdev$lambda.min), 1.4, paste("lambda.min=",round(fit_regL_CVdev$lambda.min,4),"\n",length(fit_regL_coef@i), " genes" ,sep=""), col="orange", cex=0.75, pos=4) ### Plot lambda fit #plot(fit_regL_CVdev, xvar="lambda", main="Lasso coefficient") #abline(v=log(fit_regL_CVdev$lambda.min), col="orange", lwd=2) ############### #summary(cv.fit) #print(cv.fit) #cv.fit$lambda.min #cv.fit$lambda.1se #coef(cv.fit, s = "lambda.min") #----------candidates #coef.min= coef(cv.fit, s = "lambda.min") select.varialbes = rownames(as.data.frame(which(coef(cv.fit, s = "lambda.min")[,1]!=0))) #select.varialbes = select.varialbes[-1] #remove intercept #select.varialbes Select.Variables<-c(Select.Variables,select.varialbes) } if (length(Select.Variables) != 0) { LASSO.Feature <- data.frame(Feature=Select.Variables,Cancer=project) %>% group_by(Cancer,Feature)%>% summarise(Count=n()) %>% rbind.data.frame(LASSO.Feature) } } #write.csv(LASSO.Feature,file = "Risk.LASSO.COX.csv",row.names = F) ######## Risk score Confirm Risk group Cancer #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(survival) library(survminer) LASSO.Feature <- read.csv("Risk.LASSO.COX.csv") Feature.Count <- LASSO.Feature %>% filter(Count >= 50) %>% group_by(Cancer) %>% summarise(Count=n()) %>% filter(Count >= 2) Feature.Filter <- LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) CIndex <- data.frame() RiskScore.Cancer <- data.frame() files <- list.files("RiskGroup",pattern = "Risk.Model.rds") Projects <- files %>% str_remove_all("\\..*") dir.create("RiskGroup") for (project in Projects) { print(project) phenotype <- read.csv(paste("../../Clinical.XENA/",project,"/","TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) #Confirm.Feature <- Feature.Filter %>% # filter(Cancer==project) #### survival datasets survival.data <- read.csv(paste("../../Clinical.XENA/",project,"/TCGA-",project,".survival.tsv",sep = ""),sep = "\t",header = T,check.names = F) Inter.Samples <- intersect(intersect(phenotype$submitter_id.samples,colnames(TPM)),survival.data$sample) #### symbol expression #TPM.ENSEMBL.Target <- TPM.log[Confirm.Feature$Feature,Inter.Samples] %>% # t() %>% data.frame(check.names = F) %>% # mutate(SampleID = rownames(.)) %>% # merge(.,survival.data,by.x="SampleID",by.y="sample") ## formaula #formula <- as.formula(paste0("Surv(OS.time,OS)~",paste(Confirm.Feature$Feature,collapse = "+"))) #multi_variate_cox <- coxph(formula,data=TPM.ENSEMBL.Target) #saveRDS(multi_variate_cox,file = file.path("RiskGroup",paste(project,".Risk.Model.rds",sep = ""))) ## check PH hypothesis #ph_hypo_multi <- cox.zph(multi_variate_cox) #ph_hypo_table <- ph_hypo_multi$table[-nrow(ph_hypo_multi$table),] multi_variate_cox <- readRDS(file = file.path("RiskGroup",paste(project,".Risk.Model.rds",sep = ""))) TPM.ENSEMBL.Target <- TPM.log[names(multi_variate_cox$coefficients),Inter.Samples] %>% t() %>% data.frame(check.names = F) %>% mutate(SampleID = rownames(.)) %>% merge(.,survival.data,by.x="SampleID",by.y="sample") #riskscore <- function(survival_cancer_df,candidate_genes_for_cox,cox_model){ # library(dplyr) # risk_score_table <- survival_cancer_df %>% # dplyr::select(candidate_genes_for_cox) # for (each_sig_gene in colnames(risk_score_table)) { # risk_score_table$each_sig_gene <- risk_score_table[,each_sig_gene]*(summary(cox_model)$coefficients[each_sig_gene,1]) # } # risk_score_table <- cbind(risk_score_table, 'total_risk_score'=exp(rowSums(risk_score_table))) %>% # cbind(survival_cancer_df[,c("SampleID",'OS.time','OS')]) # risk_score_table <- risk_score_table[,c("SampleID",'OS.time','OS',candidate_genes_for_cox,'total_risk_score')] # return(risk_score_table) #} RiskScore.Cancer <- predict(multi_variate_cox, type="risk", newdata=TPM.ENSEMBL.Target[,names(multi_variate_cox$coefficients)]) %>% as.data.frame() %>% magrittr::set_colnames("total_risk_score") %>% #mutate(TPM.ENSEMBL.Target$SampleID) %>% cbind.data.frame(TPM.ENSEMBL.Target) %>% data.frame(check.names = F) %>% dplyr::select(c("SampleID",'OS.time','OS',"total_risk_score")) %>% mutate(Cancer=project)%>% rbind.data.frame(RiskScore.Cancer) COX.Summary <- summary(multi_variate_cox) CIndex <- data.frame(C.Index = COX.Summary$concordance[1]) %>% mutate(Cancer=project) %>% rbind.data.frame(CIndex) } #write.csv(CIndex,file = "RiskGroup.COX.TCGA.CIndex.csv",row.names = F) #write.csv(RiskScore.Cancer,file = "RiskGroup.COX.TCGA.RiskScore.csv",row.names = F) #### Figure => C index + Gene count => 23 Cancers #### BCAA.ID <- read.csv("BCAA.ID.csv") LASSO.Feature <- read.csv("Risk.LASSO.COX.csv") CIndex <- read.csv(file = "RiskGroup.COX.TCGA.CIndex.csv") %>% #arrange(C.Index) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) Immune.Col <- c("#293462","#a64942","#fe5f55","#fff1c1", "#5bd1d7","#348498","#004d61","#ff502f", "#e41749","#f5587b","#ff8a5c","#fff591", "#001871","#ff585d","#ffb549","#41b6e6", "#515bd4","#8134af","#dd2a7b","#feda77", "#96ceb4","#ffeead","#d9534f","#ffad60", "#05445c","#f2317f","#5c4f74","#040000", "#de4307","#f29c2b","#f6d04d","#8bc24c", "#fef4a9","#3b9a9c","#4bc2c5","#78fee0", "#fad3cf","#a696c8","#2470a0","#c50d66", "#f07810","#eec60a","#dd7777","#1687a7", "#014955","#dd0a35","#ED5485","#FFE869", "#bc8420","#F0B775","#D25565","#2E94B9") p1.B <- ggplot(CIndex,aes(x=C.Index,y=Cancer,fill=Cancer))+ geom_segment(aes(x=0.5,xend=C.Index,y=Cancer,yend=Cancer))+ geom_point(shape=21,color="black",size=3)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="C index",y="")+ scale_fill_manual(values = Immune.Col)+ scale_x_continuous(limits = c(0.5,1))+ geom_vline(xintercept = seq(0.5,0.9,0.2),linetype=2) #ggthemes::scale_color_stata() Feature.Count <- LASSO.Feature %>% filter(Count >= 50) %>% group_by(Cancer) %>% summarise(Count=n()) %>% filter(Count >= 2) %>% mutate(Cancer=factor(Cancer,levels = levels(CIndex$Cancer))) library(RColorBrewer) Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[1:3], brewer.pal(8,"Dark2")[6], brewer.pal(12,"Set3")[4:12], brewer.pal(8,"Dark2")[1:4]) p1.C <- ggplot(Feature.Count,aes(Count,Cancer,fill=Cancer))+ geom_segment(aes(x=0,xend=Count,y=Cancer,yend=Cancer))+ geom_point(shape=21,color="black",size=3)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="Gene count",y="")+ #ggthemes::scale_color_stata() scale_fill_manual(values = Immune.Col)+ scale_x_continuous(limits = c(0,16)) p1 <- p1.B %>% aplot::insert_right(p1.C) ggsave(p1,filename = "RiskGroup.COX.TCGA.CIndex.Figure.pdf",height = 5,width = 3.2) p1.B %>% aplot::insert_right(p1.C) %>% aplot::insert_right(p2) p<-LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) %>% merge(BCAA.ID,by.x="Feature",by.y="ENSEMBL") %>% mutate(Count=1) %>% #mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% ggplot(aes(Cancer,Count,fill=SYMBOL))+ geom_col(color="black")+ #geom_text(aes(label=Count),hjust=1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_blank(), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",fill="")+ #ggthemes::scale_color_stata() scale_fill_manual(values = Cancer.Col)#+ #scale_x_continuous(limits = c(0,20)) ggsave(p,filename = "RiskGroup.COX.TCGA.CIndex.CancerWithGene.pdf",width = 3.5,height = 4.8) ggsave(p,filename = "RiskGroup.COX.TCGA.CIndex.CancerWithGene2.pdf",width = 7,height = 4) dataset.Sankey <- LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) %>% merge(BCAA.ID,by.x="Feature",by.y="ENSEMBL") %>% mutate(Count=1) write.csv(dataset.Sankey,file = "dataset.Sankey.csv",row.names = F) dataset.Sankey <- read.csv("dataset.Sankey.csv") library(sankeywheel) sankeywheel(from = dataset.Sankey$Cancer, to = dataset.Sankey$SYMBOL, weight = dataset.Sankey$Count, title = "", type = "dependencywheel", theme = "sunset", width = "200%") Data.Circle <- dataset.Sankey %>% dplyr::select(Cancer,SYMBOL,Count) %>% spread(Cancer,Count,fill=0) %>% remove_rownames() %>% column_to_rownames("SYMBOL") %>% as.matrix() library(circlize) circos.par(gap.after = c(rep(5, nrow(mat)-1), 15, rep(5, ncol(mat)-1), 15), start.degree = 85) chordDiagram(Data.Circle, order = c(rev(colnames(Data.Circle)), rev(rownames(Data.Circle)))) circos.clear() Feature.Filter <- LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) %>% merge(BCAA.ID,by.x="Feature",by.y="ENSEMBL") %>% group_by(SYMBOL) %>% summarise(Count=n()) %>% arrange(Count) %>% mutate(SYMBOL=factor(SYMBOL,levels = c(.$SYMBOL))) p2 <- ggplot(Feature.Filter,aes(Count,SYMBOL,fill=SYMBOL))+ geom_col()+ geom_text(aes(label=Count),hjust=1)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="No. of cancer",y="")+ #ggthemes::scale_color_stata() scale_fill_manual(values = Immune.Col)+ scale_x_continuous(limits = c(0,20)) ggsave(p2,filename = "RiskGroup.COX.TCGA.CIndex.GeneForCancer.pdf",width = 5,height = 4) #### Figure => TimeROC => 23 Cancers ###### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(survminer) library(survivalROC) library(survival) dir.create("riskScore.timeROC") AUC.Data <- data.frame() RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") for (project in unique(RiskScore.Cancer$Cancer)) { middata <- RiskScore.Cancer %>% filter(Cancer==project) library(timeROC) with(middata, ROC_riskscore <<- timeROC(T = OS.time, delta = OS, marker = total_risk_score, cause = 1, weighting = "marginal", times = c(365,1080,1800), ROC = TRUE, iid = TRUE) ) #plot(ROC_riskscore, time = 365, col = "red", add = F,title = "") #plot(ROC_riskscore, time = 1080, col = "blue", add = T) #plot(ROC_riskscore, time = 1800, col = "purple", add = T) #legend("bottomright",c("1-Year","3-Year","5-Year"),col=c("red","blue","purple"),lty=1,lwd=2) #text(0.5,0.2,paste("1-Year AUC = ",round(ROC_riskscore$AUC[1],3))) #text(0.5,0.15,paste("3-Year AUC = ",round(ROC_riskscore$AUC[2],3))) #text(0.5,0.1,paste("5-Year AUC = ",round(ROC_riskscore$AUC[3],3))) dat = data.frame(fpr = as.numeric(ROC_riskscore$FP), tpr = as.numeric(ROC_riskscore$TP), time = rep(as.factor(c(365,1095,1825)),each = nrow(ROC_riskscore$TP))) write.csv(dat,file = file.path("riskScore.timeROC",paste("ROC.Data.",project,".csv",sep = "")),row.names = F) AUC.Data <- data.frame(year1=round(ROC_riskscore$AUC,2)[1], year3=round(ROC_riskscore$AUC,2)[2], year5=round(ROC_riskscore$AUC,2)[3]) %>% mutate(Cancer=project) %>% rbind.data.frame(AUC.Data) library(ggplot2) p <- ggplot() + geom_line(data = dat,aes(x = fpr, y = tpr,color = time),size = 1) + scale_color_manual(name = NULL,values = c("#E64B35FF", "#3C5488FF", "#00A087FF"), labels = paste0("AUC of ",c(1,3,5),"-y survival: ", format(round(ROC_riskscore$AUC,2),nsmall = 2)))+ #ggsci::scale_color_npg()+ geom_line(aes(x=c(0,1),y=c(0,1)),color = "grey",linetype=2)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x = "1 - Specificity", y = "Sensitivity")+ coord_fixed() ggsave(p,filename = file.path("riskScore.timeROC",paste("ROC.",project,".pdf",sep = "")), height = 3,width = 5) #result <-with(new_dat, #survivalROC(Stime=time, # status=event, # marker=fp, # predict.time=365, # method="KM")) } write.csv(AUC.Data,file = file.path("riskScore.timeROC","AUC.Pancancer.csv"),row.names = F) #### Figure #### AUC.Data <- read.csv(file.path("riskScore.timeROC","AUC.Pancancer.csv")) %>% gather(Survival,AUC,-Cancer) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) Cancer.Col <- c(brewer.pal(9,"Set1")[1:5], brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[3:3], brewer.pal(8,"Dark2")[6], brewer.pal(12,"Set3")[4:6], brewer.pal(12,"Set3")[8:12], brewer.pal(8,"Dark2")[1:4]) p2 <- ggplot(AUC.Data,aes(x=AUC,y=reorder(Cancer,AUC),fill=Cancer))+ geom_segment(aes(x=0,y=Cancer,xend=AUC,yend=Cancer))+ facet_wrap(~Survival,ncol = 3)+ geom_point(shape=21,color="black",size=3)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x = "AUROC", y = "")+ scale_fill_manual(values = Immune.Col)+ geom_vline(xintercept = seq(0.5,0.9,0.2),linetype=2) p2.1 <- p1.B %>% aplot::insert_right(p1.C) %>% aplot::insert_right(p2,width = 3) ggsave(p2.1,filename = "RiskGroup.COX.TCGA.timeROC.Figure.pdf",height = 5,width = 7) ######## Validation ###### #### PCAWG => linux ##### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(survival) library(survminer) LASSO.Feature <- read.csv("Risk.LASSO.COX.csv") BCAA.ID <- read.csv("BCAA.ID.csv") Feature.Count <- LASSO.Feature %>% filter(Count >= 50) %>% group_by(Cancer) %>% summarise(Count=n()) %>% filter(Count >= 2) Feature.Filter <- LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) library(readxl) PCAWG <- read_xlsx("PCAWG/pcawg_donor_clinical_August2016_v9.xlsx") Express <- read.csv("PCAWG/pcawg.RNA.TPM.csv") Specimen <- read.csv("PCAWG/pcawg.sp_specimen_type.txt",sep = "\t") %>% filter(str_detect(dcc_specimen_type,"Primary")) Survival.D <- read.csv("PCAWG/pcawg.survival_sp.txt",sep = "\t") Sample.Project <- read.csv("PCAWG/pcawg.project_code_sp.txt",sep = "\t") %>% filter(!str_detect(dcc_project_code,"-US")) %>% filter(dcc_project_code %in% c("OV-AU","PRAD-CA","PRAD-UK", "LIRI-JP","PACA-AU","PACA-CA", "LAML-KR","ESAD-UK","LINC-JP", "LICA-FR","EOPC-DE","BRCA-UK", "BRCA-EU","BTCA-SG","GACA-CN")) %>% merge(data.frame(dcc_project_code = c("OV-AU","PRAD-CA","PRAD-UK", "LIRI-JP","PACA-AU","PACA-CA", "LAML-KR","ESAD-UK","LINC-JP", "LICA-FR","EOPC-DE","BRCA-UK", "BRCA-EU","BTCA-SG","GACA-CN"), Cancer = c("OV","PRAD","PRAD", "LIHC","PAAD","PAAD", "LAML","ESCA","LIHC", "LIHC","PRAD","BRCA","BRCA", "CHOL","STAD"))) Phenotype <- merge(Specimen,Survival.D,by.x="icgc_specimen_id",by.y="xena_sample") %>% merge(Sample.Project,by = "icgc_specimen_id") for (project1 in unique(Phenotype$dcc_project_code)) { print(project1) Target.Data <- Phenotype %>% filter(dcc_project_code == project1) project2 <- Target.Data$Cancer %>% unique() if (project2 %in% Feature.Filter$Cancer) { middata <- Feature.Filter %>% filter(Cancer == project2) Exp <- Express %>% filter(X %in% middata$Feature) %>% remove_rownames() %>% column_to_rownames("X") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("icgc_specimen_id") %>% merge(Target.Data,by="icgc_specimen_id") print(nrow(Exp)) if (nrow(Exp) >= 0) { saveRDS(Exp,file = paste("BCAA.riskScore.",project1,".",project2,".RDS",sep = "")) } } } ###### Validation => Function #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(survival) library(survminer) LASSO.Feature <- read.csv("Risk.LASSO.COX.csv") BCAA.ID <- read.csv("BCAA.ID.csv") Feature.Count <- LASSO.Feature %>% filter(Count >= 50) %>% group_by(Cancer) %>% summarise(Count=n()) %>% filter(Count >= 2) Feature.Filter <- LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) %>% merge(BCAA.ID,by.x = "Feature",by.y = "ENSEMBL") Cal_function <- function(dataset,feature,project,timeVector){ return.list <- list() formula <- as.formula(paste0("Surv(OS.time,OS)~", paste(feature,collapse = "+"))) dataset$OS <- as.numeric(as.double(dataset$OS)) dataset$OS.time <- as.numeric(as.double(dataset$OS.time)) multi_variate_cox <- coxph(formula,data=dataset) dataset$total_risk_score <- predict(multi_variate_cox, type="risk", newdata=dataset[,feature]) return.list[[1]] = dataset COX.Summary <- summary(multi_variate_cox) return.list[[2]] = COX.Summary$concordance[1] library(timeROC) with(dataset, ROC_riskscore <<- timeROC(T = OS.time, delta = OS, marker = total_risk_score, cause = 1, weighting = "marginal", times = timeVector,# c(12,36,60), ROC = TRUE, iid = TRUE) ) return.list[[3]] = ROC_riskscore res.cut <- surv_cutpoint(dataset, time = paste("OS",".time",sep = ""), event = "OS", variables = "total_risk_score") dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(Surv(OS.time, OS) ~ total_risk_score, data = dat) index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","total_risk_score",sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) return.list[[4]] = pValue library(ggthemes) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) p<-ggsurvplot(fit,pval =TRUE, data = dat, surv.median.line = "hv", legend.title = "Risk", conf.int.style = "step", xlab = "Time in days", #break.time.by = 500, risk.table = "abs_pct", #risk.table.y.text.col = T, #risk.table.y.text = FALSE, legend.labs = c("High", "Low"), #pval = TRUE, conf.int = TRUE, palette = "Set1", #ggtheme = theme.sur, risk.table.y.text.col = T, risk.table.y.text = F, ggtheme = theme.sur) print(p) #dev.off() library(export) graph2pdf(file=paste("RiskGroup/Validation.OS.",project,".RiskScore.BestCutoff.KM.pdf",sep = ''),height = 5,width = 4) names(return.list) <- c("Dataset","C.index","timeROC","KM.Pvalue") return(return.list) } #### LIHC => LIRI-JP #### LIHC.Feature <- Feature.Filter %>% filter(Cancer == "LIHC") Exp.Samples <- read.csv("../../../../HepG2-CBX2/Validate/ICGC-LIRI-JP/HCCDB18.sample.txt",sep = "\t",check.names = F,row.names = 1) %>% t() %>% data.frame(check.names = F) %>% filter(TYPE=="HCC") %>% rownames_to_column("SampleID") %>% dplyr::select(SampleID,TYPE,PATIENT_ID) Exp.Clinical <- read.csv("../../../../HepG2-CBX2/Validate/ICGC-LIRI-JP/HCCDB18.patient.txt",sep = "\t",check.names = F,row.names = 1) %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("PATIENT_ID") %>% mutate(OS=if_else(STATUS=="Dead",1,0)) %>% mutate(OS.time=SUR) %>% dplyr::select(PATIENT_ID,PATIENT,OS,OS.time) %>% merge(Exp.Samples,by="PATIENT_ID") Exp <- read.csv("../../../../HepG2-CBX2/Validate/ICGC-LIRI-JP/HCCDB18_mRNA_level3.txt",sep = "\t",check.names = F) Exp.Data <- Exp[,-1] %>% data.frame(check.names = F) %>% remove_rownames() %>% filter(Symbol %in% LIHC.Feature$SYMBOL) %>% column_to_rownames("Symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("SampleID") %>% merge(Exp.Clinical,by="SampleID") DATA <- Cal_function(Exp.Data,LIHC.Feature$SYMBOL,"LIHC.LIRI-JP",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.LIHC.LIRI-JP.RDS") #### COAD => GSE28722 #### library(data.table) COAD.Feature <- Feature.Filter %>% filter(Cancer == "COAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE28722.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% COAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE28722.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,COAD.Feature$SYMBOL,"COAD.GSE28722",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.COAD.GSE28722.RDS") #### COAD => GSE103479 #### Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE103479.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% COAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE103479.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,COAD.Feature$SYMBOL,"COAD.GSE103479",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.COAD.GSE103479.RDS") #### LUSC => GSE157009 #### LUSC.Feature <- Feature.Filter %>% filter(Cancer == "LUSC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE157009.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LUSC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE157009.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LUSC.Feature$SYMBOL,"LUSC.GSE157009",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LUSC.GSE157009.RDS") #### LUSC => GSE37745 #### LUSC.Feature <- Feature.Filter %>% filter(Cancer == "LUSC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE37745.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LUSC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE37745.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LUSC.Feature$SYMBOL,"LUSC.GSE37745",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LUSC.GSE37745.RDS") #### LUSC => GSE50081 #### LUSC.Feature <- Feature.Filter %>% filter(Cancer == "LUSC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE50081.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LUSC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE50081.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LUSC.Feature$SYMBOL,"LUSC.GSE50081",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LUSC.GSE50081.RDS") #### BLCA => GSE31684 #### BLCA.Feature <- Feature.Filter %>% filter(Cancer == "BLCA") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE31684.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% BLCA.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE31684.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,BLCA.Feature$SYMBOL,"BLCA.GSE31684",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.BLCA.GSE31684.RDS") #### BLCA => GSE69795 #### BLCA.Feature <- Feature.Filter %>% filter(Cancer == "BLCA") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE69795.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% BLCA.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE69795.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,BLCA.Feature$SYMBOL,"BLCA.GSE69795",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.BLCA.GSE69795.RDS") #### ESCA => GSE53625 #### ESCA.Feature <- Feature.Filter %>% filter(Cancer == "ESCA") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE53625.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% ESCA.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE53625.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,ESCA.Feature$SYMBOL,"ESCA.GSE53625",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.ESCA.GSE53625.RDS") #### ESCA => GSE53624 #### ESCA.Feature <- Feature.Filter %>% filter(Cancer == "ESCA") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE53624.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% ESCA.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE53624.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,ESCA.Feature$SYMBOL,"ESCA.GSE53624",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.ESCA.GSE53624.RDS") #### HNSC => GSE65858 #### HNSC.Feature <- Feature.Filter %>% filter(Cancer == "HNSC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE65858.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% HNSC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE65858.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,HNSC.Feature$SYMBOL,"HNSC.GSE65858",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.HNSC.GSE65858.RDS") #### KIRC => GSE167573 #### KIRC.Feature <- Feature.Filter %>% filter(Cancer == "KIRC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE167573.gene_expression_RNAseq.Normalized_Counts.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% KIRC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE167573.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("SampleID"),by="SampleID") Exp.Data <- na.omit(Exp.Data) DATA <- Cal_function(Exp.Data,KIRC.Feature$SYMBOL,"KIRC.GSE167573",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.KIRC.GSE167573.RDS") #### LAML => GSE71014 #### LAML.Feature <- Feature.Filter %>% filter(Cancer == "LAML") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE71014.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LAML.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE71014.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LAML.Feature$SYMBOL,"LAML.GSE71014",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LAML.GSE71014.RDS") #### LAML => GSE106291 #### LAML.Feature <- Feature.Filter %>% filter(Cancer == "LAML") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE106291.gene_expression_RNAseq.Counts.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LAML.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE106291.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% mutate(SampleID=str_remove_all(SampleID,"-")) %>% merge(Exp.Data %>% rownames_to_column("SampleID") %>% mutate(SampleID=str_remove_all(SampleID,"-")), by="SampleID") DATA <- Cal_function(Exp.Data,LAML.Feature$SYMBOL,"LAML.GSE106291",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LAML.GSE106291.RDS") #### LAML => GSE146173 #### LAML.Feature <- Feature.Filter %>% filter(Cancer == "LAML") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE146173.gene_expression_RNAseq.Salmon_Estimated_Counts.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LAML.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE146173.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% mutate(SampleID=str_remove_all(SampleID,"-") %>% str_remove_all(":.*")) %>% merge(Exp.Data %>% rownames_to_column("SampleID") %>% mutate(SampleID=str_remove_all(SampleID,"-")), by="SampleID") DATA <- Cal_function(Exp.Data,LAML.Feature$SYMBOL,"LAML.GSE146173",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LAML.GSE146173.RDS") #### LAML => GSE12417 #### LAML.Feature <- Feature.Filter %>% filter(Cancer == "LAML") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE12417.gene_expression_array.GPL570.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LAML.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE12417.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LAML.Feature$SYMBOL,"LAML.GSE12417",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LAML.GSE12417.RDS") #### LGG => GSE121720 #### LGG.Feature <- Feature.Filter %>% filter(Cancer == "LGG") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE121720.gene_expression_RNAseq.TPM.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LGG.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE121720.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient," .*")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("SampleID"),by="SampleID") DATA <- Cal_function(Exp.Data,LGG.Feature$SYMBOL,"LGG.GSE121720",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LGG.GSE121720.RDS") #### LGG => GSE4412 #### LGG.Feature <- Feature.Filter %>% filter(Cancer == "LGG") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE4412.gene_expression_array.GPL96.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LGG.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE4412.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LGG.Feature$SYMBOL,"LGG.GSE4412",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LGG.GSE4412.RDS") #### LGG => GSE7696 #### LGG.Feature <- Feature.Filter %>% filter(Cancer == "LGG") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE7696.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LGG.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE7696.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LGG.Feature$SYMBOL,"LGG.GSE7696",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LGG.GSE7696.RDS") #### LIHC => GSE76427 #### LIHC.Feature <- Feature.Filter %>% filter(Cancer == "LIHC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE76427.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LIHC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE76427.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LIHC.Feature$SYMBOL,"LIHC.GSE76427",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LIHC.GSE76427.RDS") #### LIHC => GSE116174 #### LIHC.Feature <- Feature.Filter %>% filter(Cancer == "LIHC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE116174.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LIHC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE116174.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LIHC.Feature$SYMBOL,"LIHC.GSE116174",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LIHC.GSE116174.RDS") #### LIHC => GSE144269 #### LIHC.Feature <- Feature.Filter %>% filter(Cancer == "LIHC") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE144269.gene_expression_RNAseq.Counts.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LIHC.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE144269.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*\\[") %>% str_remove_all("\\]")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("SampleID"),by="SampleID") DATA <- Cal_function(Exp.Data,LIHC.Feature$SYMBOL,"LIHC.GSE144269",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LIHC.GSE144269.RDS") #### LUAD => GSE42127 #### LUAD.Feature <- Feature.Filter %>% filter(Cancer == "LUAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE42127.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% LUAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE42127.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,LUAD.Feature$SYMBOL,"LUAD.GSE42127",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LUAD.GSE42127.RDS") #### LUAD => GSE30219 #### LUAD.Feature <- Feature.Filter %>% filter(Cancer == "LUAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE30219_expression.txt",header = T) %>% data.frame(check.names = F) %>% filter(`Gene Symbol` %in% LUAD.Feature$SYMBOL) %>% column_to_rownames("Gene Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE30219_clinical.txt",header = T) %>% data.frame(check.names = F) %>% mutate(OS=if_else(status=="DEAD",1,0), OS.time=`disease free survival in months`) %>% filter(source=="Lung Tumour") %>% filter(OS.time!="na") %>% filter(OS!="na") %>% dplyr::select(accession,OS.time,OS) %>% merge(Exp.Data %>% rownames_to_column("accession"),by="accession") DATA <- Cal_function(Exp.Data,LUAD.Feature$SYMBOL,"LUAD.GSE30219",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.LUAD.GSE30219.RDS") #### LUAD => GSE31210 #### LUAD.Feature <- Feature.Filter %>% filter(Cancer == "LUAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE31210_expression.txt",header = T) %>% data.frame(check.names = F) %>% filter(`Gene Symbol` %in% LUAD.Feature$SYMBOL) %>% column_to_rownames("Gene Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE31210_clinical.txt",header = T) %>% data.frame(check.names = F) %>% mutate(OS=if_else(death=="dead",1,0), OS.time=`days before death/censor`) %>% #filter(source=="Lung Tumour") %>% filter(OS.time!="na") %>% filter(OS!="na") %>% dplyr::select(accession,OS.time,OS) %>% filter(OS.time!="na") %>% filter(OS!="na") %>% mutate(OS=as.numeric(as.character(OS))) %>% mutate(OS.time=as.numeric(as.character(OS.time))) %>% merge(Exp.Data %>% rownames_to_column("accession"),by="accession") DATA <- Cal_function(Exp.Data,LUAD.Feature$SYMBOL,"LUAD.GSE31210",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.LUAD.GSE31210.RDS") #### MESO => GSE17118 #### MESO.Feature <- Feature.Filter %>% filter(Cancer == "MESO") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE17118.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% MESO.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE17118.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,MESO.Feature$SYMBOL,"MESO.GSE17118",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.MESO.GSE17118.RDS") #### OV => GSE138866 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE138866.gene_expression_RNAseq.DESeq2_Normalized_Counts.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE138866.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("SampleID"),by="SampleID") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE138866",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE138866.RDS") #### OV => GSE63885 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE63885.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE63885.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE63885",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE63885.RDS") #### OV => GSE17260 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE17260.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE17260.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE17260",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE17260.RDS") #### OV => GSE73614 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE73614.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE73614.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE73614",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE73614.RDS") #### OV => GSE26712 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE26712.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE26712.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE26712",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE26712.RDS") #### OV => GSE18520 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE18520.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE18520.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE18520",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE18520.RDS") #### OV => GSE30161 #### OV.Feature <- Feature.Filter %>% filter(Cancer == "OV") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE30161.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% OV.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE30161.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,OV.Feature$SYMBOL,"OV.GSE30161",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.OV.GSE30161.RDS") #### PAAD => GSE62452 #### PAAD.Feature <- Feature.Filter %>% filter(Cancer == "PAAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE62452.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% PAAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE62452.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,PAAD.Feature$SYMBOL,"PAAD.GSE62452",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.PAAD.GSE62452.RDS") #### PAAD => GSE28735 #### PAAD.Feature <- Feature.Filter %>% filter(Cancer == "PAAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE28735.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% PAAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE28735.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,PAAD.Feature$SYMBOL,"PAAD.GSE28735",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.PAAD.GSE28735.RDS") #### PAAD => GSE85916 #### PAAD.Feature <- Feature.Filter %>% filter(Cancer == "PAAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE85916.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% PAAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE85916.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,PAAD.Feature$SYMBOL,"PAAD.GSE85916",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.PAAD.GSE85916.RDS") #### PAAD => GSE79668 #### PAAD.Feature <- Feature.Filter %>% filter(Cancer == "PAAD") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE79668.gene_expression_RNAseq.Counts.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% PAAD.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE79668.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,"_SL.*")%>%str_remove_all("_T$")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(SampleID,OS_Time,OS_Status) %>% magrittr::set_colnames(c("SampleID","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("SampleID"),by="SampleID") DATA <- Cal_function(Exp.Data,PAAD.Feature$SYMBOL,"PAAD.GSE79668",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.PAAD.GSE79668.RDS") #### PRAD => GSE107299 #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/CPC-Gene_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- log2(Exp.Data@assayData$exprs+1) %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") Exp.Data <- Exp.Data %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(Sur.Data,by="sample_id") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(sample_type == "Primary") DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.GSE107299",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.GSE107299.RDS") #### PRAD => GSE21034#### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/Taylor_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- log2(Exp.Data@assayData$exprs+1) %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") Exp.Data <- Exp.Data %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(sample_type == "Primary") DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.GSE21034",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.GSE21034.RDS") #### PRAD => DKFZ #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/DKFZ_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- log2(Exp.Data@assayData$exprs+1) %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(sample_type == "Primary") DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.DKFZ",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.DKFZ.RDS") #### PRAD => GSE54460 #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/GSE54460_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- Exp.Data@assayData$exprs %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") Exp.Data <- Exp.Data %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(sample_type == "Primary") DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.GSE54460",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.GSE54460.RDS") #### PRAD => GSE70768 #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/Cambridge_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- Exp.Data@assayData$exprs %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(sample_type == "Primary") DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.GSE70768",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.GSE70768.RDS") #### PRAD => GSE70769 #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/Stockholm_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- Exp.Data@assayData$exprs %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(sample_type == "Primary") DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.GSE70769",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.GSE70769.RDS") #### PRAD => GSE70767 #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/CancerMap_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- Exp.Data@assayData$exprs %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(str_detect(sample_type,"Tumor")) DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.GSE70767",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.GSE70767.RDS") #### PRAD => E-MTAB-6128 #### PRAD.Feature <- Feature.Filter %>% filter(Cancer == "PRAD") Exp.Data <- readRDS("../AnimoAcidSensing/Validation.Survival/CIT_eSet.RDS") Sur.Data <- Exp.Data@phenoData@data Exp.Data <- Exp.Data@assayData$exprs %>% data.frame(check.names = F) %>% rownames_to_column("Gene_Symbol") %>% filter(Gene_Symbol %in% PRAD.Feature$Feature) %>% data.frame(check.names = F) %>% remove_rownames() %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Accession") %>% merge(Sur.Data %>% rownames_to_column("Accession"),by="Accession") %>% mutate(OS=bcr_status, OS.time=time_to_bcr) %>% filter(str_detect(sample_type,"Primary")) DATA <- Cal_function(Exp.Data,PRAD.Feature$Feature,"PRAD.E-MTAB-6128",c(12,36,60)) saveRDS(DATA,file = "riskScore.Validation.PRAD.E-MTAB-6128.RDS") #### READ => GSE87211 #### READ.Feature <- Feature.Filter %>% filter(Cancer == "READ") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE87211.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% READ.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE87211.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,READ.Feature$SYMBOL,"READ.GSE87211",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.READ.GSE87211.RDS") #### READ => GSE103479 #### READ.Feature <- Feature.Filter %>% filter(Cancer == "READ") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE103479.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% READ.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE103479.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,READ.Feature$SYMBOL,"READ.GSE103479",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.READ.GSE103479.RDS") #### READ => GSE12945 #### READ.Feature <- Feature.Filter %>% filter(Cancer == "READ") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE12945.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% READ.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE12945.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,READ.Feature$SYMBOL,"READ.GSE12945",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.READ.GSE12945.RDS") #### SKCM => GSE22153 #### SKCM.Feature <- Feature.Filter %>% filter(Cancer == "SKCM") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE22153.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% SKCM.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE22153.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,SKCM.Feature$SYMBOL,"SKCM.GSE22153",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.SKCM.GSE22153.RDS") #### SKCM => GSE65904 #### SKCM.Feature <- Feature.Filter %>% filter(Cancer == "SKCM") Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE65904_expression.txt",header = T) %>% data.frame(check.names = F) %>% filter(Symbol %in% SKCM.Feature$SYMBOL) %>% column_to_rownames("Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE65904_clinical.txt",header = T) %>% data.frame(check.names = F) %>% mutate(OS=`disease specific survival (1=death, 0=alive)`, OS.time=`disease specific survival in days`) %>% #mutate(SampleID=str_remove_all(Patient,".*, ")) %>% #filter(Sample_Type == "Tumor") %>% dplyr::select(accession,OS.time,OS) %>% #magrittr::set_colnames(c("accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("accession"),by="accession") DATA <- Cal_function(Exp.Data,SKCM.Feature$SYMBOL,"SKCM.GSE65904",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.SKCM.GSE65904.RDS") #### SKCM => GSE19234 #### SKCM.Feature <- Feature.Filter %>% filter(Cancer == "SKCM") library(data.table) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE19234.gene_expression_array.tsv.gz",header = T) %>% data.frame(check.names = F) %>% filter(Gene_Symbol %in% SKCM.Feature$SYMBOL) %>% column_to_rownames("Gene_Symbol") %>% t() %>% data.frame(check.names = F) #Exp.Data <- log2(Exp.Data+1) Exp.Data <- fread("../AnimoAcidSensing/Validation.Survival/GSE19234.clinical.tsv.gz",header = T) %>% data.frame(check.names = F) %>% mutate(SampleID=str_remove_all(Patient,".*, ")) %>% filter(Sample_Type == "Tumor") %>% dplyr::select(Accession,OS_Time,OS_Status) %>% magrittr::set_colnames(c("Accession","OS.time","OS")) %>% merge(Exp.Data %>% rownames_to_column("Accession"),by="Accession") DATA <- Cal_function(Exp.Data,SKCM.Feature$SYMBOL,"SKCM.GSE19234",c(365,1080,1800)) saveRDS(DATA,file = "riskScore.Validation.SKCM.GSE19234.RDS") ##### Figure ##### files <- list.files(".",pattern = "^riskScore") files <- files[str_detect(files,"RDS$")] Cindex <- data.frame() AUC.Data <- data.frame() KM.P <- data.frame() for (file in files) { project <- (file %>% str_split("\\.") %>% unlist())[3] code <- (file %>% str_split("\\.") %>% unlist())[4] print(code) data <- readRDS(file) AUC.Data <- data.frame(year1=round(data$timeROC$AUC,2)[1], year3=round(data$timeROC$AUC,2)[2], year5=round(data$timeROC$AUC,2)[3]) %>% mutate(Cancer=project, Code=code) %>% rbind.data.frame(AUC.Data) Cindex <- data.frame(C.Index = data$C.index) %>% mutate(Cancer=project, Code=code) %>% rbind.data.frame(Cindex) KM.P <- data.frame(KM.Pvalue = data$KM.Pvalue) %>% mutate(Cancer=project, Code=code) %>% rbind.data.frame(KM.P) dat = data.frame(fpr = as.numeric(data$timeROC$FP), tpr = as.numeric(data$timeROC$TP), time = rep(as.factor(data$timeROC$times), each = nrow(data$timeROC$TP))) library(ggplot2) p <- ggplot() + geom_line(data = dat,aes(x = fpr, y = tpr,color = time),size = 1) + scale_color_manual(name = NULL,values = c("#E64B35FF", "#3C5488FF", "#00A087FF"), labels = paste0("AUC of ",c(1,3,5),"-y survival: ", format(round(data$timeROC$AUC,2),nsmall = 2)))+ #ggsci::scale_color_npg()+ geom_line(aes(x=c(0,1),y=c(0,1)),color = "grey",linetype=2)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x = "1 - Specificity", y = "Sensitivity")+ coord_fixed() ggsave(p,filename = file.path("riskScore.timeROC",paste("ROC.",project,"_",code,".pdf",sep = "")), height = 3,width = 5) } library(RColorBrewer) Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[1:3], brewer.pal(8,"Dark2")[6], brewer.pal(12,"Set3")[4:12], brewer.pal(8,"Dark2")[1:4]) Cindex <- Cindex %>% arrange(desc(Cancer),Code) %>% mutate(Code=factor(Code,levels = .$Code)) middata <- Cindex %>% group_by(Cancer) %>% summarise(Count=n()) %>% arrange(desc(Cancer)) RECT<-data.frame(xmax=accumulate(middata$Count,sum)+0.5, xmin=accumulate(middata$Count,sum)-middata$Count+0.5, Cancer=middata$Cancer) write.csv(Cindex,file = "riskScore.Validation.Cindex.csv",row.names = F) P1 <- ggplot(data = Cindex,aes(C.Index,Code,fill=Cancer))+ geom_segment(aes(x=0.5,xend=C.Index, y=Code,yend=Code), color="black")+ geom_point(size=3,shape=21,color="black")+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x = "C index", y = "")+ annotate("rect", xmin=-Inf,xmax=Inf, ymin=RECT$xmin, ymax=RECT$xmax, fill=rev(Cancer.Col[1:16]), alpha = 0.4 )+ scale_fill_manual(values = Cancer.Col)+ scale_color_manual(values = Cancer.Col)+ geom_vline(xintercept = seq(0.6,1,0.2),linetype=2) P1.R <- ggplot(Cindex,aes(1,Code,fill=Cancer))+ geom_col(width = 1)+ theme_void()+ #theme(legend.position = "bottom")+ scale_fill_manual(values = Cancer.Col) library(tidyverse) AUC.Data <- AUC.Data %>% gather(Year,AUROC,year1:year5) #%>% #arrange(desc(Cancer),Code) #%>% #mutate(Code=factor(Code,levels = levels(Cindex$Code))) write.csv(AUC.Data,file = "riskScore.Validation.AUROC.csv",row.names = F) P2 <- ggplot(AUC.Data,aes(AUROC,Code,fill=Cancer))+ geom_segment(aes(x=0.2, xend=AUROC, y=Code, yend=Code),color="black")+ geom_point(size=3,shape=21,color="black")+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x = "AUROC", y = "")+ facet_wrap(~Year,ncol = 3)+ scale_fill_manual(values = Cancer.Col)+ scale_color_manual(values = Cancer.Col)+ geom_vline(xintercept = seq(0.6,1,0.2),linetype=2) KM.P <- KM.P %>% #gather(Year,AUROC,year1:year5) %>% arrange(desc(Cancer),Code) %>% mutate(Code=factor(Code,levels = levels(Cindex$Code))) P3 <- ggplot()+ geom_segment(data=KM.P,aes(x=0,xend=-log10(KM.Pvalue),y=Code,yend=Code), size=1,color="black")+ geom_point(data=KM.P, aes(x=-log10(KM.Pvalue),y=Code, fill=-log10(KM.Pvalue),size=-log10(KM.Pvalue)), color="white",stroke=1,shape=21)+ geom_point(data=KM.P %>% filter(KM.Pvalue <= 0.05), aes(x=-log10(KM.Pvalue),y=Code, fill=-log10(KM.Pvalue),size=-log10(KM.Pvalue)), color="black",stroke=1,shape=21)+ scale_size_continuous(range = c(1,5))+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x = "-log10(P)", y = "")+ annotate("rect", xmin=-Inf,xmax=Inf, ymin=RECT$xmin, ymax=RECT$xmax, fill=rev(Cancer.Col[1:16]), alpha = 0.4 )+ scale_x_continuous(limits = c(0,14))+ geom_vline(xintercept = c(-log10(0.05)),linetype=2)+ #scale_color_manual(values = Cancer.Col)+ scale_fill_gradientn(colours = brewer.pal(11,"RdBu")[5:1]) P.F <- P1 %>% aplot::insert_left(P1.R,width = 0.1) %>% aplot::insert_right(P2,width = 3) %>% aplot::insert_right(P3,width = 1) ggsave(P.F,filename = "riskScore.Validation.pdf",height = 8,width = 12) ggsave(P.F,filename = "riskScore.Validation.Legend.pdf",height = 8,width = 12) ######## Risk score Risk Group => Survival #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") Risk.Cutoff <- data.frame() Survival.data <- data.frame() for (project in unique(RiskScore.Cancer$Cancer)) { mid.Risk <- RiskScore.Cancer %>% filter(Cancer == project) res.cut <- surv_cutpoint(mid.Risk, time = paste("OS",".time",sep = ""), event = "OS", variables = "total_risk_score") #res.cat <- surv_categorize(res.cut) Risk.Cutoff <- data.frame(CutPoint=res.cut$cutpoint[1]) %>% mutate(Cancer=project) %>% rbind.data.frame(Risk.Cutoff) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(Surv(OS.time, OS) ~ total_risk_score, data = dat) library(ggthemes) p<-ggsurvplot(fit,pval =TRUE, data = dat, surv.median.line = "hv", legend.title = "Risk", conf.int.style = "step", xlab = "Time in days", #break.time.by = 500, risk.table = "abs_pct", #risk.table.y.text.col = T, #risk.table.y.text = FALSE, legend.labs = c("High", "Low"), #pval = TRUE, conf.int = TRUE, palette = "Set1", #ggtheme = theme.sur, risk.table.y.text.col = T, risk.table.y.text = F, ggtheme = theme.sur) print(p) #dev.off() library(export) graph2pdf(file=paste("RiskGroup/OS.",project,".RiskScore.BestCutoff.KM.pdf",sep = ''),height = 5,width = 4) index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","total_risk_score",sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) cox <- coxph(as.formula(paste("Surv(OS.time, OS) ~ ","total_risk_score",sep = "")), data = mid.Risk) coxSummary = summary(cox) Survival.data=rbind(Survival.data, data.frame(ENSEMBL="RiskScore", KM.pvalue=pValue, HR=coxSummary$conf.int[,"exp(coef)"], HR.95L=coxSummary$conf.int[,"lower .95"], HR.95H=coxSummary$conf.int[,"upper .95"], HR.pvalue=coxSummary$coefficients[,"Pr(>|z|)"], Cancer=project)) } write.csv(Survival.data,file = "RiskGroup.COX.TCGA.RiskScore.Survival.csv",row.names = F) write.csv(Risk.Cutoff,file = "RiskGroup.COX.TCGA.RiskScore.CutOff.csv",row.names = F) ######## Risk score => TIDE #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") TIDE.data <- read.csv("../../TIDE/TIDE.csv") Risk.Cutoff <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.CutOff.csv") RiskScore.TIDE <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% filter(SampleID %in% PhenoType$Sample.ID) %>% mutate(Sample.ID = substring(SampleID,1,12)) %>% merge(TIDE.data,by.x="Sample.ID",by.y="SampleID") %>% #merge(Risk.Cutoff,by.x = "Cancer.x",by.y="Cancer") %>% group_by(Cancer.x) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) write.csv(RiskScore.TIDE,file = "RiskGroup.COX.TCGA.RiskScore.TIDE.csv",row.names = F) library(ggpubr) data2 <- compare_means(formula = Exclusion~Group,data = RiskScore.TIDE,method = "wilcox.test", group.by="Cancer.x") write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.TIDE.Exclusion.Wilcox.csv") data3 <- compare_means(formula = Dysfunction~Group,data = RiskScore.TIDE,method = "wilcox.test", group.by="Cancer.x") write.csv(data3,file = "RiskGroup.COX.TCGA.RiskScore.TIDE.Dysfunction.Wilcox.csv") library(RColorBrewer) p1<-ggplot()+ geom_boxplot(data=RiskScore.TIDE, aes(x=Exclusion,y=reorder(Cancer.x,Exclusion), #Cancer,# fill=Group))+ geom_text(data=data2,aes(x=0.3,y=Cancer.x,label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_x_continuous(limits = c(-0.3,0.4))+ ggtitle("Exclusion")+ theme(plot.title = element_text(hjust = 0.5)) p2<-ggplot()+ geom_boxplot(data=RiskScore.TIDE, aes(x=Dysfunction,y=reorder(Cancer.x,Dysfunction), #Cancer,# fill=Group))+ geom_text(data=data3,aes(x=0.5,y=Cancer.x,label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_x_continuous(limits = c(-0.4,0.6))+ ggtitle("Dysfunction")+ theme(plot.title = element_text(hjust = 0.5)) p12 <- p1 %>% aplot::insert_right(p2,width = 1) ggsave(p12,filename = "RiskGroup.COX.TCGA.RiskScore.TIDE.pdf",height = 4,width = 5) ######## Risk score => IMMUNECELL #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") TIMER2 <- read.csv("../../TIMER2/infiltration_estimation_for_tcga.csv", check.names = F) %>% remove_rownames() %>% column_to_rownames("cell_type") %>% dplyr::select(ends_with("CIBERSORT")) %>% rownames_to_column("Sample.ID") %>% gather(CellType,Abundance,-Sample.ID) %>% mutate(CellType=str_remove_all(CellType,"_CIBERSORT")) Risk.Cutoff <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.CutOff.csv") RiskScore.TIMER2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% filter(SampleID %in% PhenoType$Sample.ID) %>% mutate(Sample.ID = substring(SampleID,1,15)) %>% merge(TIMER2,by="Sample.ID") %>% #merge(Risk.Cutoff,by = "Cancer") %>% group_by(Cancer,CellType) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) write.csv(RiskScore.TIMER2,file = "RiskGroup.COX.TCGA.RiskScore.CIBERSORT.csv",row.names = F) data2 <- compare_means(formula = Abundance~Group,data = RiskScore.TIMER2,method = "wilcox.test", group.by=c("Cancer","CellType")) write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.CIBERSORT.Wilcox.csv",row.names = F) RiskScore.TIMER2 <- read.csv("RiskGroup.COX.TCGA.RiskScore.CIBERSORT.csv") data2 <- read.csv("RiskGroup.COX.TCGA.RiskScore.CIBERSORT.Wilcox.csv") RiskScore.TIMER2.FigureD <- RiskScore.TIMER2 %>% group_by(Cancer,CellType) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% group_by(Cancer,CellType,Group) %>% summarise(MeanValue=mean(Abundance)) %>% ungroup() %>% data.frame(check.names = F) %>% merge(data2 %>% dplyr::select(Cancer,CellType,p.signif), by=c("Cancer","CellType")) %>% mutate(p.signif=if_else(p.signif=="ns","",p.signif)) %>% #filter(Attribute %in% c("IPS","TMB","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss")) %>% spread(Group,MeanValue) %>% mutate(Enrich=if_else(p.signif == "","n.s.",if_else(High>Low,"High","Low"))) %>% gather(Group,Score,High:Low) %>% arrange(Cancer,CellType) %>% group_by(Cancer,CellType) %>% mutate(Enrich2=if_else(Enrich=="n.s.","n.s.",if_else(Enrich==Group,"Higher","Lower"))) %>% mutate(CellType=factor(CellType,levels = unique(sort(CellType)))) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% mutate(p.signif=factor(p.signif,levels = c("*","**","***","****",""))) %>% mutate(Xaxis = paste(CellType,Group,sep = "_")) write.csv(RiskScore.TIMER2.FigureD,file = "RiskGroup.COX.TCGA.RiskScore.CIBERSORT.FigureData.csv",row.names = F) p1 <- ggplot(RiskScore.TIMER2.FigureD,aes(x=Xaxis,y=Cancer))+ geom_point(aes(size=Score,color=p.signif,shape=Enrich2), stroke=1.2)+ scale_size_continuous("Mean",range = c(2,4))+ #facet_wrap(~Attribute,ncol = 11)+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_blank(), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("P value",values = c(brewer.pal(9,"Set1")[1:4],"#F7F7F7"))+ scale_shape_manual("Score",values = c(1,3,4))+ geom_vline(xintercept = seq(0.5,44,2),linetype=2) p2 <- ggplot(RiskScore.TIMER2.FigureD %>% data.frame() %>% dplyr::select(Xaxis,Group) %>% unique(), aes(x=Xaxis,y=1,fill=Group))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("Risk score",values = c("#fe5f55","#004d61")) p3 <- ggplot(RiskScore.TIMER2.FigureD %>% data.frame() %>% dplyr::select(Xaxis,CellType) %>% unique(), aes(x=Xaxis,y=1,fill=CellType))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("Cell type",values = Cancer.Col) p123 <- p1 %>% aplot::insert_top(p2,height = 0.05) %>% aplot::insert_top(p3,height = 0.05) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.CIBERSORT.FigureData.pdf",height = 4,width = 12) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.CIBERSORT.FigureData.Legend.pdf",height = 16,width = 12) for (project in unique(RiskScore.TIMER2$Cancer)) { middata <- RiskScore.TIMER2 %>% filter(Cancer==project) Label.Data <- data2 %>% filter(Cancer==project) p <- ggplot()+ geom_boxplot(data=middata, aes(x=Abundance,y=CellType, #Cancer,# fill=Group))+ geom_text(data=Label.Data, aes(x=max(middata$Abundance)+0.03,y=CellType, label=p.signif))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Phenotype,scale="free_y",ncol = 6)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_x_continuous(limits = c(0,max(middata$Abundance)+0.1)) ggsave(p,filename = paste("RiskGroup/RiskGroup.COX.",project,".RiskScore.CIBERSORT.pdf",sep = ""),height = 5,width = 5) } #scale_x_continuous(trans = "log2") #scale_x_continuous(limits = c(-0.3,0.4))+ #ggtitle("Exclusion")+ #theme(plot.title = element_text(hjust = 0.5)) ######## Risk score => INDEX #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% mutate(ID=substring(SampleID,1,15)) PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") Risk.Index <- data.frame() data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 3) data2<-data[,-1] Risk.Index <- merge(RiskScore.Cancer,data2,by="ID") %>% unique() %>% data.frame() %>% filter(SampleID %in% PhenoType$Sample.ID) %>% gather(Attribute,Score,MHC:IPS) %>% dplyr::select(ID,SampleID,Attribute,Score,Cancer,total_risk_score) %>% rbind.data.frame(Risk.Index) for (index in c(4,6,7,10,5,8,9,13)) { data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = index) data2<-data[,-1] Risk.Index <- merge(RiskScore.Cancer,data2,by.x="ID",by.y="SampleName") %>% unique() %>% data.frame() %>% filter(SampleID %in% PhenoType$Sample.ID) %>% mutate(Attribute=colnames(data2)[1]) %>% mutate(Score=.[[colnames(data2)[1]]]) %>% dplyr::select(ID,SampleID,Attribute,Score,Cancer,total_risk_score) %>% rbind.data.frame(Risk.Index) } Risk.Index2 <- Risk.Index %>% group_by(Cancer,Attribute) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) write.csv(Risk.Index,file = "RiskGroup.COX.TCGA.RiskScore.Index.csv",row.names = F) data2 <- compare_means(formula = Score~Group,data = Risk.Index2,method = "wilcox.test", group.by=c("Cancer","Attribute")) write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.Index.Wilcox.csv",row.names = F) #### Figure #### Risk.Index <- read.csv("RiskGroup.COX.TCGA.RiskScore.Index.csv") data2 <- read.csv("RiskGroup.COX.TCGA.RiskScore.Index.Wilcox.csv") Risk.Index2 <- Risk.Index %>% group_by(Cancer,Attribute) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% group_by(Cancer,Attribute,Group) %>% summarise(MeanValue=mean(Score)) %>% ungroup() %>% data.frame(check.names = F) %>% merge(data2 %>% dplyr::select(Cancer,Attribute,p.signif), by=c("Cancer","Attribute")) %>% mutate(p.signif=if_else(p.signif=="ns","",p.signif)) %>% filter(Attribute %in% c("IPS","TMB","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss")) %>% spread(Group,MeanValue) %>% mutate(Enrich=if_else(p.signif == "","n.s.",if_else(High>Low,"High","Low"))) %>% gather(Group,Score,High:Low) %>% arrange(Cancer,Attribute) %>% group_by(Cancer,Attribute) %>% mutate(Enrich2=if_else(Enrich=="n.s.","n.s.",if_else(Enrich==Group,"Higher","Lower"))) #mutate(Attribute=factor(Attribute,levels = c("IPS","TMB","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss"))) %>% #mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% #mutate(p.signif=factor(p.signif,levels = c("*","**","***","****",""))) library(ggpubr) RiskScore.TIDE <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.TIDE.csv") %>% gather(Attribute,Score,Exclusion:Dysfunction) %>% dplyr::select(Cancer.x,Attribute,Score,Group) %>% magrittr::set_colnames(c("Cancer","Attribute","Score","Group")) data2 <- compare_means(formula = Score~Group,data = RiskScore.TIDE,method = "wilcox.test", group.by=c("Cancer","Attribute")) Risk.Index3 <- RiskScore.TIDE %>% #group_by(Cancer,Attribute) %>% #mutate(cutpoint=median(total_risk_score)) %>% #mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% group_by(Cancer,Attribute,Group) %>% summarise(MeanValue=mean(Score)) %>% ungroup() %>% data.frame(check.names = F) %>% merge(data2 %>% dplyr::select(Cancer,Attribute,p.signif),by=c("Cancer","Attribute")) %>% mutate(p.signif=if_else(p.signif=="ns","",p.signif)) %>% #filter(Attribute %in% c("IPS","TMB","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss")) %>% spread(Group,MeanValue) %>% mutate(Enrich=if_else(p.signif == "","n.s.",if_else(High>Low,"High","Low"))) %>% gather(Group,Score,High:Low) %>% arrange(Cancer,Attribute) %>% group_by(Cancer,Attribute) %>% mutate(Enrich2=if_else(Enrich=="n.s.","n.s.",if_else(Enrich==Group,"Higher","Lower"))) Risk.Index4 <- rbind(Risk.Index2,Risk.Index3) %>% data.frame(check.names = F) %>% mutate(Attribute=factor(Attribute,levels = c("IPS","TMB","Exclusion","Dysfunction","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss"))) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% mutate(p.signif=factor(p.signif,levels = c("*","**","***","****",""))) p <- ggplot(Risk.Index4,aes(x=Group,y=Cancer))+ geom_point(aes(size=Score,color=p.signif,shape=Enrich2))+ scale_size_continuous("Mean",range = c(2,5))+ facet_wrap(~Attribute,ncol = 11)+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("P value",values = c(brewer.pal(9,"Set1")[1:4],"grey"))+ scale_shape_manual("Score",values = c(15:17)) #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)]) ggsave(p,filename = "RiskGroup.COX.TCGA.RiskScore.Index.All.pdf",height = 4,width = 10) for (attr in unique(Risk.Index2$Attribute)) { middata <- Risk.Index2 %>% filter(Attribute == attr) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) data3 <- data2 %>% filter(Attribute == attr) p1 <- ggplot()+ geom_boxplot(data=middata, aes(x=Score,y=Cancer, #Cancer,# fill=Group))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])#+ #scale_x_continuous(trans = "log2") p2 <- ggplot(data3,aes(x=0.5,y=Cancer))+ geom_text(aes(x=0.5,y=Cancer,label=p.signif),vjust=0.5)+ theme_void() p12 <- p1 %>% aplot::insert_right(p2,width = 0.1) ggsave(p12,filename = paste("RiskGroup.COX.TCGA.RiskScore.Index.",attr,".pdf",sep = ""), height = 4.5,width = 4) } ######## Risk score => ESTIMATE #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% mutate(ID=substring(SampleID,1,15)) PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") library(readxl) data <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 2) data2<-data[,-1] Index.data2 <- merge(RiskScore.Cancer,data2,by="ID") %>% unique() %>% data.frame() %>% filter(SampleID %in% PhenoType$Sample.ID) %>% gather(Attribute,Score,StromalScore:ESTIMATEScore) %>% dplyr::select(ID,SampleID,Attribute,Score,Cancer,total_risk_score) %>% group_by(Cancer,Attribute) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) write.csv(Index.data2,file = "RiskGroup.COX.TCGA.RiskScore.ESTIMATE.csv",row.names = F) data2 <- compare_means(formula = Score~Group,data = Index.data2,method = "wilcox.test", group.by=c("Cancer","Attribute")) write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.ESTIMATE.Wilcox.csv",row.names = F) #### Figure #### for (attr in unique(Index.data2$Attribute)) { middata <- Index.data2 %>% filter(Attribute == attr) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) data3 <- data2 %>% filter(Attribute == attr) p1 <- ggplot()+ geom_boxplot(data=middata, aes(x=Score,y=Cancer, #Cancer,# fill=Group))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])#+ #scale_x_continuous(trans = "log2") p2 <- ggplot(data3,aes(x=0.5,y=Cancer))+ geom_text(aes(x=0.5,y=Cancer,label=p.signif),vjust=0.5)+ theme_void() p12 <- p1 %>% aplot::insert_right(p2,width = 0.1) ggsave(p12,filename = paste("RiskGroup.COX.TCGA.RiskScore.ESTIMATE.",attr,".pdf",sep = ""), height = 4.5,width = 4.5) } ######## Risk score => Immune score ##### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") TIMER2 <- read.csv("../../TIMER2/infiltration_estimation_for_tcga.csv", check.names = F) %>% remove_rownames() %>% column_to_rownames("cell_type") %>% dplyr::select(contains("score")) %>% rownames_to_column("Sample.ID") %>% gather(CellType,Abundance,-Sample.ID) RiskScore.TIMER2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% filter(SampleID %in% PhenoType$Sample.ID) %>% mutate(Sample.ID = substring(SampleID,1,15)) %>% merge(TIMER2,by="Sample.ID") %>% #merge(Risk.Cutoff,by = "Cancer") %>% group_by(Cancer,CellType) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() %>% filter(!str_detect(CellType,"cytotoxicity")) %>% mutate(CellType=str_remove_all(CellType,"_.*")) write.csv(RiskScore.TIMER2,file = "RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.csv",row.names = F) data2 <- compare_means(formula = Abundance~Group,data = RiskScore.TIMER2,method = "wilcox.test", group.by=c("Cancer","CellType")) write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.Wilcox.csv",row.names = F) data2 <- read.csv("RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.Wilcox.csv") data2 %>% filter(Cancer=="BLCA") #### Figure #### for (attr in unique(RiskScore.TIMER2$CellType)) { middata <- RiskScore.TIMER2 %>% filter(CellType == attr) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) data3 <- data2 %>% filter(CellType == attr) p1 <- ggplot()+ geom_boxplot(data=middata, aes(x=Abundance,y=Cancer, #Cancer,# fill=Group))+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=SampleType),size=1,dodge.width=0.75,priority = "ascending")+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ ggtitle(Hmisc::capitalize(attr))+ theme(plot.title = element_text(hjust = 0.5)) #scale_x_continuous(trans = "log2") p2 <- ggplot(data3,aes(x=0.5,y=Cancer))+ geom_text(aes(x=0.5,y=Cancer,label=p.signif),vjust=0.5)+ theme_void() p12 <- p1 %>% aplot::insert_right(p2,width = 0.1) ggsave(p12,filename = paste("RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.",Hmisc::capitalize(attr),".pdf",sep = ""), height = 4.5,width = 4.5) } #### BLCA #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") TIMER2 <- read.csv("../../TIMER2/infiltration_estimation_for_tcga.csv", check.names = F) %>% remove_rownames() %>% column_to_rownames("cell_type") %>% dplyr::select(contains("score")) %>% rownames_to_column("Sample.ID") %>% gather(CellType,Abundance,-Sample.ID) RiskScore.TIMER2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% filter(SampleID %in% PhenoType$Sample.ID) %>% mutate(Sample.ID = substring(SampleID,1,15)) %>% merge(TIMER2,by="Sample.ID") %>% #merge(Risk.Cutoff,by = "Cancer") %>% group_by(Cancer,CellType) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() %>% filter(!str_detect(CellType,"cytotoxicity")) %>% mutate(CellType=str_remove_all(CellType,"_.*")) %>% filter(Cancer=="BLCA") %>% mutate(Group=factor(Group,levels = c("Low","High"))) library(ggbeeswarm) p12 <- ggplot(RiskScore.TIMER2,aes(x=Abundance,y=Group,fill=Group))+ geom_violin(width=1.4) + geom_boxplot(width=0.2, color="grey", alpha=1)+ #geom_jitter()+ #stat_ellipse(aes(fill=Group),type = "norm", geom ="polygon",alpha=0.2,color=NA)+ #geom_beeswarm(aes(color=Group),size=1,dodge.width=0.75,priority = "ascending")+ facet_grid(CellType~.)+ #stat_compare_means(aes(group=CellType),comparisons = list(c("Low","High")))+ labs(x="Score",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(2:1)])#+ #scale_x_continuous(trans = "log2") #ggtitle(Hmisc::capitalize(attr))+ #theme(plot.title = element_text(hjust = 0.5)) ggsave(p12,filename = "RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.BLCA.pdf",height = 4,width = 2) ######## Risk score => Immune score+ESTIMATE ###### RiskScore.TIMER2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.csv") data2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.ImmuneScore.XCELL.Wilcox.csv") Wilcox.Data <- data2 %>% dplyr::select(Cancer,CellType,group1,group2,p.signif) %>% gather(Name,Group,group1:group2) %>% dplyr::select(-Name) RiskScore.TIMER2.FigureD1 <- RiskScore.TIMER2 %>% group_by(Cancer,CellType) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% group_by(Cancer,CellType,Group) %>% summarise(MeanValue=mean(Abundance)) %>% ungroup() %>% data.frame(check.names = F) %>% merge(data2 %>% dplyr::select(Cancer,CellType,p.signif), by=c("Cancer","CellType")) %>% mutate(p.signif=if_else(p.signif=="ns","",p.signif)) %>% #filter(Attribute %in% c("IPS","TMB","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss")) %>% spread(Group,MeanValue) %>% mutate(Enrich=if_else(p.signif == "","n.s.",if_else(High>Low,"High","Low"))) %>% gather(Group,Score,High:Low) %>% arrange(Cancer,CellType) %>% group_by(Cancer,CellType) %>% mutate(Enrich2=if_else(Enrich=="n.s.","n.s.",if_else(Enrich==Group,"Higher","Lower"))) colnames(RiskScore.TIMER2.FigureD1)[2]="Attribute" Index.data2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.ESTIMATE.csv") data2 <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.ESTIMATE.Wilcox.csv") RiskScore.TIMER2.FigureD2 <- Index.data2 %>% group_by(Cancer,Attribute) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% group_by(Cancer,Attribute,Group) %>% summarise(MeanValue=mean(Score)) %>% ungroup() %>% data.frame(check.names = F) %>% merge(data2 %>% dplyr::select(Cancer,Attribute,p.signif), by=c("Cancer","Attribute")) %>% mutate(p.signif=if_else(p.signif=="ns","",p.signif)) %>% #filter(Attribute %in% c("IPS","TMB","Neoantigen","MSI","HRD","MATH","Ploidy","Purity","RNAss")) %>% spread(Group,MeanValue) %>% mutate(Enrich=if_else(p.signif == "","n.s.",if_else(High>Low,"High","Low"))) %>% gather(Group,Score,High:Low) %>% arrange(Cancer,Attribute) %>% group_by(Cancer,Attribute) %>% mutate(Enrich2=if_else(Enrich=="n.s.","n.s.",if_else(Enrich==Group,"Higher","Lower"))) Risk.Index4 <- rbind(RiskScore.TIMER2.FigureD1,RiskScore.TIMER2.FigureD2) %>% data.frame(check.names = F) %>% mutate(Attribute=factor(Attribute, levels = c("ImmuneScore","StromalScore","ESTIMATEScore", "immune score","stroma score","microenvironment score"))) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% mutate(p.signif=factor(p.signif,levels = c("*","**","***","****",""))) p <- ggplot(Risk.Index4,aes(x=Group,y=Cancer))+ geom_point(aes(size=Score,color=p.signif,shape=Enrich2))+ scale_size_continuous("Mean",range = c(2,5))+ facet_wrap(~Attribute,ncol = 11)+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("P value",values = c(brewer.pal(9,"Set1")[1:4],"grey"))+ scale_shape_manual("Score",values = c(15:17)) #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)]) ggsave(p,filename = "RiskGroup.COX.TCGA.RiskScore.Estimate.All.pdf",height = 4.5,width = 7) ######## Risk score => HALLMARK #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") TERM <- c("HALLMARK_E2F_TARGETS", "HALLMARK_G2M_CHECKPOINT", "HALLMARK_MYC_TARGETS_V2", "HALLMARK_MYC_TARGETS_V1", "HALLMARK_MITOTIC_SPINDLE", "HALLMARK_NOTCH_SIGNALING", "HALLMARK_WNT_BETA_CATENIN_SIGNALING", "HALLMARK_MTORC1_SIGNALING", "HALLMARK_TNFA_SIGNALING_VIA_NFKB", "HALLMARK_TGF_BETA_SIGNALING", "HALLMARK_KRAS_SIGNALING_UP", "HALLMARK_KRAS_SIGNALING_DN", "HALLMARK_IL6_JAK_STAT3_SIGNALING", "HALLMARK_INTERFERON_GAMMA_RESPONSE", "HALLMARK_INTERFERON_ALPHA_RESPONSE", "HALLMARK_INFLAMMATORY_RESPONSE", "HALLMARK_HEDGEHOG_SIGNALING", "HALLMARK_COMPLEMENT", "HALLMARK_DNA_REPAIR", "HALLMARK_HYPOXIA", "HALLMARK_APOPTOSIS", "HALLMARK_P53_PATHWAY", "HALLMARK_PI3K_AKT_MTOR_SIGNALING", "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION") files <- list.files("../../HALLMARK.ssGSEA.Pancancer/") RiskScore.Hallmark <- data.frame() for (file in files) { project <- file %>% str_remove_all("\\..*") %>% str_remove_all("TCGA-") print(project) data <- read.csv(file.path("../../HALLMARK.ssGSEA.Pancancer/",file), check.names = F,row.names = 1) %>% rownames_to_column("HALLMARK") %>% gather(SampleID,Score,-HALLMARK) RiskScore.Hallmark <- RiskScore.Cancer %>% data.frame(check.names = F) %>% filter(Cancer==project) %>% mutate(Group=ifelse(total_risk_score>=median(.$total_risk_score),"High","Low")) %>% merge(data,by="SampleID") %>% group_by(HALLMARK,Group) %>% #mutate(Scale.Score=scale(Score)) %>% mutate(MeanValue=mean(Score), MedianValue=median(Score), CVValue=sd(Score)/MeanValue*100) %>% ungroup() %>% filter(HALLMARK %in% TERM) %>% rbind.data.frame(RiskScore.Hallmark) } write.csv(RiskScore.Hallmark,file = "RiskGroup.COX.TCGA.RiskScore.HALLMARKscore.csv",row.names = F) data2 <- compare_means(formula = Score~Group,data = RiskScore.Hallmark,method = "wilcox.test", group.by=c("Cancer","HALLMARK")) write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.HALLMARKscore.Wilcox.csv",row.names = F) #### Figure #### Wilcox.Data <- data2 %>% dplyr::select(Cancer,HALLMARK,group1,group2,p.signif) %>% gather(Name,Group,group1:group2) %>% dplyr::select(-Name) %>% mutate(HALLMARK=str_remove(HALLMARK,"HALLMARK_")) RiskScore.Hallmark <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.HALLMARKscore.csv") RiskScore.Hallmark2 <- RiskScore.Hallmark %>% dplyr::select(Cancer,Group,HALLMARK,MeanValue,MedianValue,CVValue) %>% arrange(Cancer,Group,HALLMARK) %>% unique() %>% mutate(Xaxis=paste(Cancer,Group,sep = "_")) %>% mutate(HALLMARK=str_remove(HALLMARK,"HALLMARK_")) %>% merge(Wilcox.Data,by=c("Cancer","HALLMARK","Group")) %>% dplyr::select(Cancer,HALLMARK,Group,MedianValue,p.signif) %>% spread(Group,MedianValue) %>% mutate(Attributes=if_else(p.signif == "ns","ns", if_else(High>Low,"High","Low"))) %>% gather(Group,MedianValue,High:Low) %>% mutate(Xaxis=paste(Cancer,Group,sep = "_")) %>% mutate(HALLMARK=factor(HALLMARK,levels = rev(sort(unique(.$HALLMARK))))) write.csv(RiskScore.Hallmark2,file = "RiskGroup.COX.TCGA.RiskScore.HALLMARKscore.FigureData.csv",row.names = F) p1 <- ggplot()+ geom_point(data = RiskScore.Hallmark2, aes(x=Xaxis,y=HALLMARK,fill=Attributes, size=MedianValue), shape=21,color="white",stroke=1.2)+ geom_point(data = RiskScore.Hallmark2 %>% filter(p.signif != "ns"), aes(x=Xaxis,y=HALLMARK,fill=Attributes, size=MedianValue), shape=21,color="black",stroke=1.2)+ ggthemes::theme_few()+ scale_size_continuous("Median of\nssGSEA score",range = c(1,5))+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_blank(), axis.text.y = element_text(size=12), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) geom_vline(aes(xintercept=c(seq(2.5,44.5,2))), color="black",linetype="dashed")+ ggsci::scale_fill_nejm()+ labs(x="",y="",fill="Higher in group") Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(8,"Dark2")[1:6], brewer.pal(12,"Set3")[11:1]) p2 <- ggplot(data = RiskScore.Hallmark2 %>% dplyr::select(Xaxis,Cancer) %>% arrange(Cancer,Xaxis) %>% unique(), aes(x=Xaxis,y=1,fill=Cancer))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual(values = Cancer.Col) #ggthemes::scale_fill_colorblind() p3 <- ggplot(data = RiskScore.Hallmark2 %>% dplyr::select(Xaxis,Group) %>% arrange(Group,Xaxis) %>% unique(), aes(x=Xaxis,y=1,fill=Group))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("Risk score",values = c("#fe5f55","#004d61")) p123 <- p1 %>% aplot::insert_top(p3,height = 0.05) %>% aplot::insert_top(p2,height = 0.05) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.HALLMARKscore.Figure.pdf",height = 5,width = 12) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.HALLMARKscore.Figure.Legend.pdf",height = 12,width = 12) ######## Risk score => KEGG pathway ####### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(GSVA) library(GSVA) library(GSEABase) library(msigdbr) library(clusterProfiler) library(org.Hs.eg.db) library(tidyverse) #gene_set <- read.gmt("../../msigdb/c2.cp.kegg.v7.5.1.symbols.gmt") gene_set <- getGmt("../../msigdb/c2.cp.kegg.v7.5.1.symbols.gmt") #geneset <- gene_set %>% split(x = .$gene, f = .$term) dirs <- list.dirs("../../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] dir.create("KEGG.ssGSEA.Pancancer") for (dir in dirs) { project <- str_remove(dir,".*/") print(project) phenotype <- read.csv(paste(dir,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) mRNA.Exp.Target <- TPM #[,intersect(phenotype$submitter_id.samples,colnames(TPM.log))] %>% data.frame(check.names = F) G_list <- read.csv("../LIHC/biomaRt.GeneID.tansfer.csv") dat.2 <- mRNA.Exp.Target %>% data.frame(check.names = F) %>% rownames_to_column("ensembl_gene_id") %>% data.frame(check.names = F) %>% merge(.,G_list[,1:2],by="ensembl_gene_id") %>% data.frame(check.names = F) %>% na.omit() %>% dplyr::select(-ensembl_gene_id) expr_mean=aggregate(.~hgnc_symbol,mean,data=dat.2) #calculate mean for same symbol dat.3 <- expr_mean %>% remove_rownames() %>% column_to_rownames("hgnc_symbol") res.ssgsea <- gsva(as.matrix(dat.3), gene_set, method = "ssgsea", kcdf = "Gaussian", min.sz = 5,abs.ranking=TRUE) write.csv(res.ssgsea,file = paste("KEGG.ssGSEA.Pancancer/TCGA-",project,".ssGSEA.csv",sep = "")) } TERM <- gene_set$term[str_detect(gene_set$term,"METABOLISM$") | str_detect(gene_set$term,"BIOSYNTHESIS$") | str_detect(gene_set$term,"DEGRADATION$") | str_detect(gene_set$term,"KEGG_CELL_CYCLE")| str_detect(gene_set$term,"KEGG_DNA_REPLICATION")| str_detect(gene_set$term,"KEGG_RNA_POLYMERASE")| str_detect(gene_set$term,"KEGG_NOTCH_SIGNALING_PATHWAY")| str_detect(gene_set$term,"KEGG_HEDGEHOG_REPLICATION")| str_detect(gene_set$term,"KEGG_BASE_EXCISION_REPAIR")| str_detect(gene_set$term,"KEGG_NUCLEOTIDE_EXCISION_REPAIR")| str_detect(gene_set$term,"KEGG_MISMATCH_REPAIR")| str_detect(gene_set$term,"KEGG_HOMOLOGOUS_RECOMBINATION")| str_detect(gene_set$term,"KEGG_MAPK_SIGNALING_PATHWAY")] %>% sort() %>% unique() %>% as.character() TERM <- TERM[TERM != "KEGG_GLYCOSYLPHOSPHATIDYLINOSITOL_GPI_ANCHOR_BIOSYNTHESIS"] setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") files <- list.files(path = "KEGG.ssGSEA.Pancancer/",pattern = "csv$") RiskScore.KEGG <- data.frame() for (file in files) { project <- file %>% str_remove_all("\\..*") %>% str_remove_all("TCGA-") print(project) mid.pheno <- PhenoType %>% filter(Cancer==project) data <- read.csv(file.path("KEGG.ssGSEA.Pancancer/",file), check.names = F,row.names = 1) %>% rownames_to_column("HALLMARK") %>% gather(SampleID,Score,-HALLMARK) %>% filter(SampleID %in% mid.pheno$Sample.ID) RiskScore.KEGG <- RiskScore.Cancer %>% data.frame(check.names = F) %>% filter(Cancer==project) %>% mutate(Group=ifelse(total_risk_score>=median(.$total_risk_score),"High","Low")) %>% merge(data,by="SampleID") %>% group_by(HALLMARK,Group) %>% #mutate(Scale.Score=scale(Score)) %>% mutate(MeanValue=mean(Score), MedianValue=median(Score), CVValue=sd(Score)/MeanValue*100) %>% ungroup() %>% filter(HALLMARK %in% TERM) %>% rbind.data.frame(RiskScore.KEGG) } write.csv(RiskScore.KEGG,file = "RiskGroup.COX.TCGA.RiskScore.KEGGscore.csv",row.names = F) library(ggpubr) data2 <- compare_means(formula = Score~Group,data = RiskScore.KEGG,method = "wilcox.test", group.by=c("Cancer","HALLMARK")) write.csv(data2,file = "RiskGroup.COX.TCGA.RiskScore.KEGGscore.Wilcox.csv",row.names = F) #### Figure #### Wilcox.Data <- data2 %>% dplyr::select(Cancer,HALLMARK,group1,group2,p.signif) %>% gather(Name,Group,group1:group2) %>% dplyr::select(-Name) %>% mutate(HALLMARK=str_remove(HALLMARK,"KEGG_")) RiskScore.KEGG <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.KEGGscore.csv") RiskScore.KEGG2 <- RiskScore.KEGG %>% dplyr::select(Cancer,Group,HALLMARK,MeanValue,MedianValue,CVValue) %>% arrange(Cancer,Group,HALLMARK) %>% unique() %>% mutate(Xaxis=paste(Cancer,Group,sep = "_")) %>% mutate(HALLMARK=str_remove(HALLMARK,"KEGG_")) %>% merge(Wilcox.Data,by=c("Cancer","HALLMARK","Group")) %>% dplyr::select(Cancer,HALLMARK,Group,MedianValue,p.signif) %>% spread(Group,MedianValue) %>% mutate(Attributes=if_else(p.signif == "ns","ns", if_else(High>Low,"High","Low"))) %>% gather(Group,MedianValue,High:Low) %>% mutate(Xaxis=paste(Cancer,Group,sep = "_")) %>% mutate(HALLMARK=factor(HALLMARK,levels = rev(sort(unique(.$HALLMARK))))) write.csv(RiskScore.KEGG2,file = "RiskGroup.COX.TCGA.RiskScore.KEGGscore.FigureData.csv",row.names = F) p1 <- ggplot()+ geom_point(data = RiskScore.KEGG2, aes(x=Xaxis,y=HALLMARK,fill=Attributes, size=MedianValue), shape=21,color="white",stroke=1.2)+ geom_point(data = RiskScore.KEGG2 %>% filter(p.signif != "ns"), aes(x=Xaxis,y=HALLMARK,fill=Attributes, size=MedianValue), shape=21,color="black",stroke=1.2)+ ggthemes::theme_few()+ scale_size_continuous("Median of\nssGSEA score",range = c(1,4))+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_blank(), axis.text.y = element_text(size=9), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) geom_vline(aes(xintercept=c(seq(2.5,44.5,2))), color="black",linetype="dashed")+ ggsci::scale_fill_nejm()+ labs(x="",y="",fill="Higher in group") Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(8,"Dark2")[1:6], brewer.pal(12,"Set3")[11:1]) p2 <- ggplot(data = RiskScore.KEGG2 %>% dplyr::select(Xaxis,Cancer) %>% arrange(Cancer,Xaxis) %>% unique(), aes(x=Xaxis,y=1,fill=Cancer))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual(values = Cancer.Col) #ggthemes::scale_fill_colorblind() p3 <- ggplot(data = RiskScore.KEGG2 %>% dplyr::select(Xaxis,Group) %>% arrange(Group,Xaxis) %>% unique(), aes(x=Xaxis,y=1,fill=Group))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("Risk score",values = c("#fe5f55","#004d61")) p123 <- p1 %>% aplot::insert_top(p3,height = 0.05) %>% aplot::insert_top(p2,height = 0.05) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.KEGGscore.Figure.pdf",height = 10,width = 12) ######## Risk score => MUTATION + CNV #### #### Mutation => Cancer ##### setwd("K:/TCGA/Anlysis/BCAA") #Driver <- read.csv("mutation_download_tab.txt",sep = "\t",check.names = F) library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") dirs <- list.dirs("../../Clinical.XENA/") dirs <- dirs[str_detect(dirs,"//")] dir.create("RiskGenome") for (dir in dirs) { project <- str_remove_all(dir,".*/") print(project) filename <- paste(project,"/TCGA-",project,".mutect2_snv.tsv",sep = "") tmp = read.csv(file.path("../../Clinical.XENA/",filename),sep = "\t",header=TRUE) colnames(tmp) =c( "Tumor_Sample_Barcode", "Hugo_Symbol", "Chromosome", "Start_Position", "End_Position", "Reference_Allele", "Tumor_Seq_Allele2", "HGVSp_Short" , 'effect' ,"Consequence", "vaf" ) tmp$Entrez_Gene_Id =1 tmp$Center ='ucsc' tmp$NCBI_Build ='GRCh38' tmp$NCBI_Build ='GRCh38' tmp$Strand ='+' tmp$Variant_Classification = tmp$effect tail(sort(table(tmp$Variant_Classification ))) tmp$Tumor_Seq_Allele1 = tmp$Reference_Allele tmp$Variant_Type = ifelse( tmp$Reference_Allele %in% c('A','C','T','G') & tmp$Tumor_Seq_Allele2 %in% c('A','C','T','G'), 'SNP','INDEL' ) tmp <- tmp %>% mutate(Variant_Classification = if_else(Variant_Classification == "inframe_insertion","In_Frame_Ins", if_else(Variant_Classification == "inframe_deletion","In_Frame_Del", if_else(Variant_Classification == "frameshift_variant","Frame_Shift_Var", if_else(Variant_Classification == "missense_variant","Missense_Mutation", if_else(Variant_Classification == "synonymous_variant","Nonsense_Mutation", if_else(Variant_Classification == "3_prime_UTR_variant" | Variant_Classification == "5_prime_UTR_variant","Prime_UTR_Var", if_else(Variant_Classification == "intron_variant","Intron_Var", if_else(str_detect(Variant_Classification,";"),"Multi_Hit", if_else(str_detect(Variant_Classification,"splice"),"Splice_Site","Multi_Hit")))))))))) #DriverGene <- Driver %>% filter(cancer_type_abbr == project) #DriverGeneList <- DriverGene$driver_gene %>% str_split(",") %>% unlist() %>% unique() if (c(project) %in% RiskScore.Cancer$Cancer) { RiskGroup <- RiskScore.Cancer %>% filter(Cancer==project) %>% mutate(Group=if_else(total_risk_score >= median(total_risk_score),"High","Low")) High.Risk <- RiskGroup %>% filter(Group=="High") Low.Risk <- RiskGroup %>% filter(Group=="Low") HighRisk.Sample <- tmp %>% filter(Tumor_Sample_Barcode %in% High.Risk$SampleID) LowRisk.Sample <- tmp %>% filter(Tumor_Sample_Barcode %in% Low.Risk$SampleID) write.csv(HighRisk.Sample,file = file.path("RiskGenome",paste(project,".HighRisk.csv",sep = "")),row.names = F) write.csv(LowRisk.Sample,file = file.path("RiskGenome",paste(project,".LowRisk.csv",sep = "")),row.names = F) library(RColorBrewer) #vc_cols = c("#fff1c1", # "#5bd1d7","#348498","#ff502f","#004d61", # "#e41749","#f5587b","#ff8a5c","#8bc24c") vc_cols <- rev(pal_npg("nrc")(9)) #ggthemes::ggthemes_data$stata$colors$schemes$economist$value[14:6] names(vc_cols) <- c("In_Frame_Ins", "Missense_Mutation", "In_Frame_Del", "Frame_Shift_Var", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site", "Multi_Hit" ) #DriverGeneList2 <- DriverGeneList[DriverGeneList %in% HighRisk.Sample$Hugo_Symbol] library(maftools) HighRisk = read.maf(maf = HighRisk.Sample, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) LowRisk = read.maf(maf = LowRisk.Sample, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) library(magrittr) laml <- read.maf(maf = tmp %>% filter(Tumor_Sample_Barcode %in% RiskGroup$SampleID), clinicalData = RiskGroup %>% dplyr::select(SampleID,Group) %>% set_colnames(c("Tumor_Sample_Barcode","Riskscore")), vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) fab.ce = clinicalEnrichment(maf = laml, clinicalFeature = 'Riskscore') #Sig.Table <- fab.ce$groupwise_comparision[p_value < 0.05] write.csv(fab.ce$groupwise_comparision,file = file.path("RiskGenome/",paste("High.vs.Low.",project,".csv",sep = "")),row.names = F) GeneList <- laml@gene.summary$Hugo_Symbol[1:15] coOncoplot(m1 = HighRisk, m2 = LowRisk, m1Name = 'High risk', m2Name = 'Low risk', genes = GeneList, #genes = removeNonMutated = F, colors = vc_cols) library(export) graph2pdf(file=file.path("RiskGenome/", paste("High.vs.Low.",project,".Oncoplot.pdf",sep = "")), height = 3.5,width = 8) } } #### Mutation => Pancancer #### setwd("K:/TCGA/Anlysis/BCAA") Driver <- read.csv("mutation_download_tab.txt",sep = "\t",check.names = F) library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") dirs <- list.dirs("../../Clinical.XENA/") dirs <- dirs[str_detect(dirs,"//")] #dir.create("RiskGenome") SNP.Data <- data.frame() SNP.Clinical <- data.frame() for (dir in dirs) { project <- str_remove_all(dir,".*/") print(project) filename <- paste(project,"/TCGA-",project,".mutect2_snv.tsv",sep = "") tmp = read.csv(file.path("../../Clinical.XENA/",filename),sep = "\t",header=TRUE) colnames(tmp) =c( "Tumor_Sample_Barcode", "Hugo_Symbol", "Chromosome", "Start_Position", "End_Position", "Reference_Allele", "Tumor_Seq_Allele2", "HGVSp_Short" , 'effect' ,"Consequence", "vaf" ) tmp$Entrez_Gene_Id =1 tmp$Center ='ucsc' tmp$NCBI_Build ='GRCh38' tmp$NCBI_Build ='GRCh38' tmp$Strand ='+' tmp$Variant_Classification = tmp$effect tail(sort(table(tmp$Variant_Classification ))) tmp$Tumor_Seq_Allele1 = tmp$Reference_Allele tmp$Variant_Type = ifelse( tmp$Reference_Allele %in% c('A','C','T','G') & tmp$Tumor_Seq_Allele2 %in% c('A','C','T','G'), 'SNP','INDEL' ) tmp <- tmp %>% mutate(Variant_Classification = if_else(Variant_Classification == "inframe_insertion","In_Frame_Ins", if_else(Variant_Classification == "inframe_deletion","In_Frame_Del", if_else(Variant_Classification == "frameshift_variant","Frame_Shift_Var", if_else(Variant_Classification == "missense_variant","Missense_Mutation", if_else(Variant_Classification == "synonymous_variant","Nonsense_Mutation", if_else(Variant_Classification == "3_prime_UTR_variant" | Variant_Classification == "5_prime_UTR_variant","Prime_UTR_Var", if_else(Variant_Classification == "intron_variant","Intron_Var", if_else(str_detect(Variant_Classification,";"),"Multi_Hit", if_else(str_detect(Variant_Classification,"splice"),"Splice_Site","Multi_Hit")))))))))) if (c(project) %in% RiskScore.Cancer$Cancer) { RiskGroup <- RiskScore.Cancer %>% filter(Cancer==project) %>% mutate(Group=if_else(total_risk_score >= median(total_risk_score),"High","Low")) SNP.Data <- tmp %>% filter(Tumor_Sample_Barcode %in% RiskGroup$SampleID) %>% rbind.data.frame(SNP.Data) SNP.Clinical <- data.frame(Tumor_Sample_Barcode=tmp$Tumor_Sample_Barcode %>% unique(), Cancer=project) %>% filter(Tumor_Sample_Barcode %in% RiskGroup$SampleID) %>% merge(RiskGroup[,c("SampleID","Group")], by.x="Tumor_Sample_Barcode", by.y="SampleID") %>% rbind.data.frame(SNP.Clinical) } } write.csv(SNP.Data,file = "RiskGenome/SNP.Data.csv",row.names = F) write.csv(SNP.Clinical,file = "RiskGenome/SNP.Clinical.csv",row.names = F) setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) SNP.Data <- read.csv(file = "RiskGenome/SNP.Data.csv") SNP.Clinical <- read.csv(file = "RiskGenome/SNP.Clinical.csv") library(ggsci) vc_cols <- rev(pal_npg("nrc")(9)) #ggthemes::ggthemes_data$stata$colors$schemes$economist$value[14:6] names(vc_cols) <- c("In_Frame_Ins", "Missense_Mutation", "In_Frame_Del", "Frame_Shift_Var", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site", "Multi_Hit" ) Immune.Col <- c("#293462","#a64942","#fe5f55","#fff1c1", "#5bd1d7","#348498","#004d61","#ff502f", "#e41749","#f5587b","#ff8a5c","#fff591", "#001871","#ff585d","#ffb549","#41b6e6", "#515bd4","#8134af","#dd2a7b","#feda77", "#96ceb4","#ffeead","#d9534f","#ffad60", "#05445c","#f2317f","#5c4f74","#040000", "#de4307","#f29c2b","#f6d04d","#8bc24c", "#fef4a9","#3b9a9c","#4bc2c5","#78fee0", "#fad3cf","#a696c8","#2470a0","#c50d66", "#f07810","#eec60a","#dd7777","#1687a7", "#014955","#dd0a35","#ED5485","#FFE869", "#bc8420","#F0B775","#D25565","#2E94B9") Cancer.Col = Immune.Col[c(11:43)] names(Cancer.Col) = unique(SNP.Clinical$Cancer) Cancer.Col <- list(Cancer=Cancer.Col) Group.Col <- Immune.Col[2:1] names(Group.Col) = unique(SNP.Clinical$Group) Group.Col <- list(Group=Group.Col) SNP.Clinical2 <- SNP.Clinical %>% arrange(Cancer,Group) High.D <- SNP.Clinical %>% filter(Group=="High") Low.D <- SNP.Clinical %>% filter(Group=="Low") library(maftools) tcga <- read.maf(maf = SNP.Data,clinicalData = SNP.Clinical, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) GeneList <- tcga@gene.summary$Hugo_Symbol[1:100] tcga.high <- read.maf(maf = SNP.Data %>% filter(Tumor_Sample_Barcode %in% High.D$Tumor_Sample_Barcode), clinicalData = High.D, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) maftools::oncoplot(maf = tcga.high,bgCol = "white", colors = vc_cols, additionalFeatureCol = "white", showTitle =F, annoBorderCol = "white", sortByAnnotation = T, drawRowBar = F, genes = GeneList, #sampleOrder = SNP.Clinical2$Tumor_Sample_Barcode, clinicalFeatures=c("Cancer")) tcga.Low <- read.maf(maf = SNP.Data %>% filter(Tumor_Sample_Barcode %in% Low.D$Tumor_Sample_Barcode), clinicalData = Low.D, vc_nonSyn=c("In_Frame_Ins", "In_Frame_Del", "Frame_Shift_Var", "Missense_Mutation", "Nonsense_Mutation", "Prime_UTR_Var", "Intron_Var", "Splice_Site" )) maftools::oncoplot(maf = tcga.Low,bgCol = "white", colors = vc_cols, additionalFeatureCol = "white", showTitle =F, annoBorderCol = "white", sortByAnnotation = T, drawRowBar = F, genes = GeneList, #sampleOrder = SNP.Clinical2$Tumor_Sample_Barcode, clinicalFeatures=c("Cancer")) #### Mutation => Fisher test => Pancancer => Figure #### High.ED <- tcga.high@gene.summary Low.ED <- tcga.Low@gene.summary Gene.Fisher <- data.frame() for (gene in GeneList) { Count1 <- (High.ED[High.ED$Hugo_Symbol == gene,])$AlteredSamples Count2 <- (Low.ED[Low.ED$Hugo_Symbol == gene,])$AlteredSamples test <- fisher.test(matrix(c(Count1,3203-Count1,Count2,3183-Count2),nrow = 2)) Gene.Fisher <- data.frame(Pvalue=test$p.value) %>% mutate(Cancer="Pooled",SYMBOL=gene) %>% rbind.data.frame(Gene.Fisher) } write.csv(Gene.Fisher,file = "RiskScore.Top10.SNP.csv") Gene.Fisher$SYMBOL <- factor(Gene.Fisher$SYMBOL,levels = Gene.Fisher$SYMBOL) p <- ggplot(Gene.Fisher,aes(x=-log10(Pvalue),y=SYMBOL,color=SYMBOL))+ geom_segment(aes(x=0,xend=-log10(Pvalue),y=SYMBOL,yend=SYMBOL))+ geom_point(size=3)+ ggthemes::theme_few()+ scale_size_continuous("Median of\nssGSEA score",range = c(1,5))+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=12), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ #ggsci::scale_color_nejm() ggthemes::scale_color_stata(scheme = "s1rcolor")+ labs(y="") ggsave(p,filename = "RiskGenome.High.vs.Low.PanCancer.pdf",height = 4,width = 2.5) #### Mutation => Fisher test => Cancer => Figure #### files <- list.files("RiskGenome/",pattern = "csv$") files <- files[str_detect(files,"vs")] Fisher.Cancer <- data.frame() for (file in files) { project=(file %>% str_split("\\.") %>% unlist())[4] Fisher.Cancer <- read.csv(file.path("RiskGenome/",file)) %>% filter(p_value <= 0.05) %>% #dplyr::select(p_value,Hugo_Symbol) %>% mutate(MutatedCount = str_remove_all(n_mutated_group1," of .*")) %>% mutate(Cancer=project) %>% rbind.data.frame(Fisher.Cancer) } Fisher.Cancer <- Fisher.Cancer %>% unique() write.csv(Fisher.Cancer,file = "RiskGenome.High.vs.Low.Cancer.csv",row.names = F) Fisher.Cancer2 <- Fisher.Cancer %>% data.frame() %>% group_by(Cancer) %>% dplyr::top_n(-3,p_value) %>% arrange(Cancer) %>% mutate(Hugo_Symbol=factor(Hugo_Symbol,levels = rev(unique(.$Hugo_Symbol)))) p2 <- ggplot(Fisher.Cancer2,aes(x=Cancer,y=Hugo_Symbol, color=-log10(p_value), size=-log10(p_value)))+ geom_point()+ ggthemes::theme_few()+ scale_size_continuous("-log10(Fisher's P)",range = c(1,5))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) #geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ #ggsci::scale_color_nejm() #ggthemes::scale_color_stata(scheme = "s1rcolor")+ scale_color_gradientn("-log10(Fisher's P)",colours = brewer.pal(11,"RdBu")[5:1])+ labs(y="",x="") ggsave(p2,filename = "RiskGenome.High.vs.Low.Cancer.pdf",height = 8,width = 5) ###### CNV ##### setwd("K:/TCGA/Anlysis/BCAA") #Driver <- read.csv("mutation_download_tab.txt",sep = "\t",check.names = F) library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") for (project in unique(RiskScore.Cancer$Cancer)) { print(project) RiskScore <- RiskScore.Cancer %>% filter(Cancer==project) %>% mutate(Group=if_else(total_risk_score > median(total_risk_score),"High","Low")) %>% dplyr::select(SampleID,Group) %>% filter(SampleID %in% PhenoType$Sample.ID) library(data.table) data <- fread(file.path("../../GSCA/",paste(project,".cnv.tsv.gz",sep = ""))) %>% dplyr::select(-entrez,-cancer_type) %>% remove_rownames() %>% column_to_rownames("symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% mutate(SampleID=substring(Sample.ID,1,16)) %>% dplyr::select(-Sample.ID) HighRisk.Data <- merge(RiskScore,data,by="SampleID") %>% data.frame(check.names = F) %>% filter(Group=="High") %>% remove_rownames() %>% column_to_rownames("SampleID") %>% dplyr::select(-Group) HighRisk.Count <- nrow(HighRisk.Data) LowRisk.Data <- merge(RiskScore,data,by="SampleID") %>% data.frame(check.names = F) %>% filter(Group=="Low") %>% remove_rownames() %>% column_to_rownames("SampleID") %>% dplyr::select(-Group) LowRisk.Count <- nrow(LowRisk.Data) Amplification.Fisher <- data.frame() Deletion.Fisher <- data.frame() for (name in colnames(HighRisk.Data)) { print(name) middata1 <- HighRisk.Data[[name]] middata1[middata1<0]=0 middata1[middata1==2]=1 middata2 <- HighRisk.Data[[name]] middata2[middata2>0]=0 middata2[middata2==-2]=-1 middata3 <- LowRisk.Data[[name]] middata3[middata3<0]=0 middata3[middata3==2]=1 middata4 <- LowRisk.Data[[name]] middata4[middata4>0]=0 middata4[middata4==-2]=-1 #LowRisk.Deletion <- data.frame(SYMBOL=name,Count1=abs(sum(middata4))) %>% # mutate(Cancer=project) %>% # mutate(Count2=LowRisk.Count-Count1) %>% # rbind.data.frame(LowRisk.Deletion) #test <- fisher.test(matrix(c(sum(middata1), # HighRisk.Count-sum(middata1), # sum(middata3), # LowRisk.Count-sum(middata3)),nrow = 2)) #Amplification.Fisher <- data.frame(SYMBOL=name, # Fisher.P=test$p.value, # OR=test$estimate) %>% # mutate(High.CNV=sum(middata1),Low.CNV=sum(middata3)) %>% # mutate(High.Sample=HighRisk.Count,Low.Sample=LowRisk.Count) %>% # mutate(Cancer=project,CNVType="Amplification") %>% # rbind.data.frame(Amplification.Fisher) test <- fisher.test(matrix(c(abs(sum(middata2)), HighRisk.Count-abs(sum(middata2)), abs(sum(middata4)), LowRisk.Count-abs(sum(middata4))),nrow = 2)) Deletion.Fisher <- data.frame(SYMBOL=name, Fisher.P=test$p.value, OR=test$estimate) %>% mutate(High.CNV=abs(sum(middata2)),Low.CNV=abs(sum(middata4))) %>% mutate(High.Sample=HighRisk.Count,Low.Sample=LowRisk.Count) %>% mutate(Cancer=project,CNVType="Deletion") %>% rbind.data.frame(Deletion.Fisher) } write.csv(Deletion.Fisher,file = file.path("RiskGenome/",paste("RiskGenome.CNV.Deletion.Fisher.",project,".csv",sep = "")),row.names = F) #write.csv(Amplification.Fisher,file = file.path("RiskGenome/",paste("RiskGenome.CNV.Amplification.Fisher.",project,".csv",sep = "")),row.names = F) } #write.csv(Deletion.Fisher,file = "RiskGenome.CNV.Deletion.Fisher.Cancer.csv",row.names = F) #write.csv(Amplification.Fisher,file = "RiskGenome.CNV.Amplification.Fisher.Cancer.csv",row.names = F) Pancancer.A.CNV <- data.frame() Pancancer.D.CNV <- data.frame() for (project in unique(RiskScore.Cancer$Cancer)) { Pancancer.D.CNV <- read.csv(file.path("RiskGenome/",paste("RiskGenome.CNV.Deletion.Fisher.",project,".csv",sep = ""))) %>% rbind.data.frame(Pancancer.D.CNV) Pancancer.A.CNV <- read.csv(file.path("RiskGenome/",paste("RiskGenome.CNV.Amplification.Fisher.",project,".csv",sep = ""))) %>% rbind.data.frame(Pancancer.A.CNV) } ### amplification ### Pancancer.A.CNV2 <- Pancancer.A.CNV %>% group_by(SYMBOL) %>% summarise(High.CNV=sum(High.CNV), Low.CNV=sum(Low.CNV), High.Sample=sum(High.Sample), Low.Sample=sum(Low.Sample)) %>% mutate(Total.CNV=High.CNV+Low.CNV) %>% arrange(desc(Total.CNV)) Pancancer.A.Fisher <- data.frame() for (gene in Pancancer.A.CNV2$SYMBOL) { print(gene) middata<-Pancancer.A.CNV2 %>% filter(SYMBOL==gene) test <- fisher.test(matrix(c(middata$High.CNV, middata$Low.CNV, middata$High.Sample-middata$High.CNV, middata$Low.Sample-middata$Low.CNV),nrow = 2)) if (test$p.value <= 0.05) { if (middata$High.CNV > middata$Low.CNV) { Pancancer.A.Fisher <- data.frame(SYMBOL=gene,Pvalue=test$p.value,OR=test$estimate, Enrich="RiskHigh") %>% rbind.data.frame(Pancancer.A.Fisher) }else{ Pancancer.A.Fisher <- data.frame(SYMBOL=gene,Pvalue=test$p.value,OR=test$estimate, Enrich="RiskLow") %>% rbind.data.frame(Pancancer.A.Fisher) } }else{ Pancancer.A.Fisher <- data.frame(SYMBOL=gene,Pvalue=test$p.value,OR=test$estimate, Enrich="NS") %>% rbind.data.frame(Pancancer.A.Fisher) } } write.csv(Pancancer.A.Fisher,file = "RiskGenome/RiskGenome.CNV.Amplification.Fisher.Pancancer.csv",row.names = F) ### Deletion ### Pancancer.D.CNV2 <- Pancancer.D.CNV %>% group_by(SYMBOL) %>% summarise(High.CNV=sum(High.CNV), Low.CNV=sum(Low.CNV), High.Sample=sum(High.Sample), Low.Sample=sum(Low.Sample)) %>% mutate(Total.CNV=High.CNV+Low.CNV) %>% arrange(desc(Total.CNV)) Pancancer.D.CNV2 <- Pancancer.D.CNV %>% group_by(SYMBOL) %>% summarise(High.CNV=sum(High.CNV), Low.CNV=sum(Low.CNV), High.Sample=sum(High.Sample), Low.Sample=sum(Low.Sample)) %>% mutate(Total.CNV=High.CNV+Low.CNV) %>% arrange(desc(Total.CNV)) Pancancer.D.Fisher <- data.frame() for (gene in Pancancer.D.CNV2$SYMBOL[]) { print(gene) middata<-Pancancer.D.CNV2 %>% filter(SYMBOL==gene) test <- fisher.test(matrix(c(middata$High.CNV, middata$Low.CNV, middata$High.Sample-middata$High.CNV, middata$Low.Sample-middata$Low.CNV),nrow = 2)) if (test$p.value <= 0.05) { if (middata$High.CNV > middata$Low.CNV) { Pancancer.D.Fisher <- data.frame(SYMBOL=gene,Pvalue=test$p.value,OR=test$estimate, Enrich="RiskHigh") %>% rbind.data.frame(Pancancer.D.Fisher) }else{ Pancancer.D.Fisher <- data.frame(SYMBOL=gene,Pvalue=test$p.value,OR=test$estimate, Enrich="RiskLow") %>% rbind.data.frame(Pancancer.D.Fisher) } }else{ Pancancer.D.Fisher <- data.frame(SYMBOL=gene,Pvalue=test$p.value,OR=test$estimate, Enrich="NS") %>% rbind.data.frame(Pancancer.D.Fisher) } } write.csv(Pancancer.D.Fisher,file = "RiskGenome/RiskGenome.CNV.Deletion.Fisher.Pancancer.csv",row.names = F) #### Figure => Pancancer #### Pancancer.A.Fisher <- read.csv(file = "RiskGenome/RiskGenome.CNV.Amplification.Fisher.Pancancer.csv") Pancancer.D.Fisher <- read.csv(file = "RiskGenome/RiskGenome.CNV.Deletion.Fisher.Pancancer.csv") Pancancer.A.D.Fisher <- Pancancer.A.Fisher %>% filter(Pvalue <= 0.05) %>% mutate(CNVType="Amplification") %>% rbind.data.frame(Pancancer.D.Fisher %>% filter(Pvalue <= 0.05) %>% mutate(CNVType="Deletion")) %>% remove_rownames() %>% arrange(Pvalue) write.csv(Pancancer.A.D.Fisher,file = "RiskGenome/RiskGenome.CNV.Deletion.Amplification.Fisher.Sig.Pancancer.csv",row.names = F) Pancancer.A.D.Fisher.FigureData <- Pancancer.A.D.Fisher %>% filter(SYMBOL %in% c("TP53","MYC","PTEN","RB1","EGFR","BRAF","APC","TTN","PIK3CA")) %>% arrange(CNVType,Pvalue) %>% mutate(SYMBOL=factor(SYMBOL,levels = .$SYMBOL)) p <- ggplot()+ geom_point(data=Pancancer.A.D.Fisher.FigureData, aes(x=CNVType,y=SYMBOL,size=-log10(Pvalue),fill=Enrich), shape=21,color="black",stroke=1)+ ggthemes::theme_few()+ scale_size_continuous("-log10(Fisher's P)",range = c(1,5))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=12), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ #ggsci::scale_color_nejm() ggthemes::scale_color_stata(scheme = "s1rcolor")+ #labs(y="") ggsci::scale_fill_nejm()+ labs(x="",y="",fill="Higher in group")+ scale_x_discrete(label=c("Amp.","Del.")) ggsave(p,filename = "RiskGenome.CNV.Deletion.Amplification.Fisher.Sig.Pancancer.pdf",height = 2.5,width = 3.5) Pancancer.A.D.Fisher.Count <- Pancancer.A.D.Fisher %>% group_by(Enrich,CNVType) %>% summarise(Count=n()) write.csv(Pancancer.A.D.Fisher.Count,file = "RiskGenome/RiskGenome.CNV.Deletion.Amplification.Fisher.Sig.Pancancer.Count.csv",row.names = F) p2 <- ggplot(Pancancer.A.D.Fisher.Count,aes(x=Enrich,y=Count,fill=CNVType))+ geom_col()+ scale_y_log10()+ ggthemes::theme_few()+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,5))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=12), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) #geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ #ggsci::scale_color_nejm() ggthemes::scale_color_stata(scheme = "s1rcolor")+ #labs(y="") ggsci::scale_fill_nejm()+ labs(x="",y="Frequency",fill="CNV type")+ scale_x_discrete(label=c("High risk","Low risk")) ggsave(p2,filename = "RiskGenome.CNV.Deletion.Amplification.Fisher.Sig.Pancancer.Count.pdf",height = 2.5,width = 3.5) #### Figure => Cancer => Amplification+Deletion => IntOGen Driver gene #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) Drivergene <- read.csv("IntOGen-DriverGenes.tsv",sep = "\t") RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") Pancancer.A.CNV <- data.frame() Pancancer.D.CNV <- data.frame() for (project in unique(RiskScore.Cancer$Cancer)) { Pancancer.D.CNV <- read.csv(file.path("RiskGenome/",paste("RiskGenome.CNV.Deletion.Fisher.",project,".csv",sep = ""))) %>% rbind.data.frame(Pancancer.D.CNV) Pancancer.A.CNV <- read.csv(file.path("RiskGenome/",paste("RiskGenome.CNV.Amplification.Fisher.",project,".csv",sep = ""))) %>% rbind.data.frame(Pancancer.A.CNV) } Cancer.A.CNV <- Pancancer.A.CNV %>% filter(Fisher.P <= 0.05) %>% mutate(Enrich=if_else(High.CNV > Low.CNV,"High","Low")) write.csv(Cancer.A.CNV,file = "RiskGenome/RiskGenome.CNV.Amplification.Fisher.Cancer.csv",row.names = F) Cancer.D.CNV <- Pancancer.D.CNV %>% filter(Fisher.P <= 0.05) %>% mutate(Enrich=if_else(High.CNV > Low.CNV,"High","Low")) write.csv(Cancer.D.CNV,file = "RiskGenome/RiskGenome.CNV.Deletion.Fisher.Cancer.csv",row.names = F) Cancer.CNV <- Cancer.A.CNV %>% rbind.data.frame(Cancer.D.CNV) %>% filter(SYMBOL %in% Drivergene$Symbol[1:60]) %>% #group_by(Cancer,CNVType) %>% #top_n(-2,Fisher.P) %>% mutate(Cancer=factor(Cancer,levels = rev(sort(unique(.$Cancer))))) %>% #arrange(Cancer,CNVType,Fisher.P) %>% mutate(SYMBOL=factor(SYMBOL,levels = Drivergene$Symbol[1:60])) library(RColorBrewer) p2 <- ggplot(Cancer.CNV,aes(x=SYMBOL,y=Cancer, color=Enrich, size=-log10(Fisher.P), shape=CNVType))+ geom_point(stroke=1.2)+ scale_shape_manual(values=c(1,4))+ ggthemes::theme_few()+ scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) #geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ ggsci::scale_color_nejm()+ #ggthemes::scale_color_stata(scheme = "s1rcolor")+ #scale_color_gradientn("-log10(Fisher's P)",colours = brewer.pal(11,"RdBu")[4:1])+ labs(y="",x="") #p2.F <- p2 %>% aplot::insert_top(p2.Top,height = 0.2) ggsave(p2,filename = "RiskGenome.CNV.Amplification.Fisher.Cancer.Examples.pdf",height = 4.5,width = 12) ######## Risk score => ImmuneGenes ########### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") ICI.P.G.E <- data.frame(Protein = c("PD-1","PD-L1","PD-L2","CTLA-4","TIGIT","LAG-3","TIM-3","ARG1","CD163","CD68","FOXP3"), SYMBOL = c("PDCD1","CD274","CD273","CTLA4","TIGIT","LAG3","HAVCR2","ARG1","CD163","CD68","FOXP3"), ENSEMBL = c("ENSG00000188389","ENSG00000120217","ENSG00000197646", "ENSG00000163599","ENSG00000181847","ENSG00000089692", "ENSG00000135077","ENSG00000118520","ENSG00000177575", "ENSG00000129226","ENSG00000049768")) dirs <- list.dirs("../../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] Exp.Risk <- data.frame() for (dir in dirs) { project <- str_remove(dir,".*/") print(project) phenotype <- read.csv(paste(dir,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) Exp.Risk <- TPM.log[ICI.P.G.E$ENSEMBL,] %>% data.frame(check.names = F) %>% rownames_to_column("ENSEMBL") %>% merge(ICI.P.G.E,by="ENSEMBL") %>% dplyr::select(-ENSEMBL,-Protein) %>% column_to_rownames("SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% filter(Sample.ID %in% PhenoType$Sample.ID) %>% mutate(Cancer=project) %>% rbind.data.frame(Exp.Risk) } Exp.Risk2 <- merge(Exp.Risk,RiskScore.Cancer,by.x="Sample.ID",by.y="SampleID") write.csv(Exp.Risk2,file = "RiskGroup.COX.TCGA.RiskScore.ImmuneGenes.csv",row.names = F) #### Figure => Correlation #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) Exp.Risk2 <- read.csv("RiskGroup.COX.TCGA.RiskScore.ImmuneGenes.csv") ### PDCD1 library(RColorBrewer) Cancer.Col <- c(brewer.pal(9,"Set1"), brewer.pal(8,"Set2")[1:7], brewer.pal(12,"Set3")[1:3], brewer.pal(8,"Dark2")[6], brewer.pal(12,"Set3")[4:12], brewer.pal(8,"Dark2")[1:4]) colors <- Cancer.Col[1:23] names(colors)=sort(unique(Exp.Risk2$Cancer.x)) Spearman.Data <- data.frame() for (project in unique(Exp.Risk2$Cancer.x)) { middata <- Exp.Risk2 %>% filter(Cancer.x == project) %>% filter(SYMBOL=="PDCD1") test <- cor.test(middata[["TPM"]], middata$total_risk_score, method = "spearman", exact = F) Spearman.Data <- data.frame(Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } write.csv(Spearman.Data,file = "RiskGroup.COX.TCGA.RiskScore.PDCD1.RiskScore.Spearman.csv",row.names = F) ### Figure suppressMessages(library(rstatix)) Spearman.Data <- Spearman.Data %>% add_significance("Pvalue") p1 <- ggplot()+ geom_segment(data = Spearman.Data, aes(x=Cancer,xend=Cancer,y=0,yend=Rho), color="black")+ geom_point(data = Spearman.Data, aes(x=reorder(Cancer,Rho),y=Rho,fill=Cancer,size=-log10(Pvalue)), shape=21,color="white",stroke=1.1)+ geom_point(data = Spearman.Data %>% filter(Pvalue <= 0.05), aes(x=reorder(Cancer,Rho),y=Rho,fill=Cancer,size=-log10(Pvalue)), shape=21,color="black",stroke=1.1)+ scale_size_continuous(range = c(2,5))+ #geom_text(data=Spearman.Data,aes(x=Cancer,y=Rho,label=Pvalue.signif))+ ggthemes::theme_few()+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) #geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ #ggsci::scale_color_nejm()+ #ggthemes::scale_color_stata(scheme = "s1rcolor")+ #scale_color_gradientn("-log10(Fisher's P)",colours = brewer.pal(11,"RdBu")[4:1])+ labs(y="Spearman'CC",x="")+ scale_fill_manual(values = colors)+ geom_hline(yintercept = c(0),linetype=2)+ scale_y_continuous(limits = c(-0.3,0.65)) ggsave(p1,filename = "RiskGroup.COX.TCGA.RiskScore.PDCD1.Spearman.pdf",height = 2,width = 7) ggsave(p1,filename = "RiskGroup.COX.TCGA.RiskScore.PDCD1.Spearman.Legend.pdf",height = 10,width = 7) ### FOXP3 Spearman.Data <- data.frame() for (project in unique(Exp.Risk2$Cancer.x)) { middata <- Exp.Risk2 %>% filter(Cancer.x == project) %>% filter(SYMBOL=="FOXP3") test <- cor.test(middata[["TPM"]], middata$total_risk_score, method = "spearman", exact = F) Spearman.Data <- data.frame(Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Spearman.Data) } write.csv(Spearman.Data,file = "RiskGroup.COX.TCGA.RiskScore.FOXP3.RiskScore.Spearman.csv",row.names = F) ### Figure suppressMessages(library(rstatix)) Spearman.Data <- Spearman.Data %>% add_significance("Pvalue") p1 <- ggplot()+ geom_segment(data = Spearman.Data, aes(x=Cancer,xend=Cancer,y=0,yend=Rho), color="black")+ geom_point(data = Spearman.Data, aes(x=reorder(Cancer,Rho),y=Rho,fill=Cancer,size=-log10(Pvalue)), shape=21,color="white",stroke=1.1)+ geom_point(data = Spearman.Data %>% filter(Pvalue <= 0.05), aes(x=reorder(Cancer,Rho),y=Rho,fill=Cancer,size=-log10(Pvalue)), shape=21,color="black",stroke=1.1)+ scale_size_continuous(range = c(2,5))+ #geom_text(data=Spearman.Data,aes(x=Cancer,y=Rho,label=Pvalue.signif))+ ggthemes::theme_few()+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)])+ #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:3)]) #geom_vline(aes(xintercept=-log10(0.05)),color="black",linetype="dashed")+ #ggsci::scale_color_nejm()+ #ggthemes::scale_color_stata(scheme = "s1rcolor")+ #scale_color_gradientn("-log10(Fisher's P)",colours = brewer.pal(11,"RdBu")[4:1])+ labs(y="Spearman'CC",x="")+ scale_fill_manual(values = colors)+ geom_hline(yintercept = c(0),linetype=2)+ scale_y_continuous(limits = c(-0.3,0.65)) ggsave(p1,filename = "RiskGroup.COX.TCGA.RiskScore.FOXP3.Spearman.pdf",height = 2,width = 7) ggsave(p1,filename = "RiskGroup.COX.TCGA.RiskScore.FOXP3.Spearman.Legend.pdf",height = 10,width = 7) #### Figure #### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) Exp.Risk2 <- read.csv("RiskGroup.COX.TCGA.RiskScore.ImmuneGenes.csv") library(ggpubr) Exp.Risk2 <- Exp.Risk2 %>% gather(SYMBOL,TPM,FOXP3:CD273) %>% group_by(Cancer.x) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() data2 <- compare_means(formula = TPM~Group,data = Exp.Risk2,method = "wilcox.test", group.by=c("Cancer.x","SYMBOL")) write.csv(data2,file="RiskGroup.COX.TCGA.RiskScore.ImmuneGenes.Wilcoxon.csv",row.names = F) Exp.Risk3 <- Exp.Risk2 %>% group_by(Cancer.x,SYMBOL,Group) %>% summarise(MeanValue=mean(TPM)) %>% ungroup() %>% data.frame(check.names = F) %>% merge(data2 %>% dplyr::select(Cancer.x,SYMBOL,p.signif), by=c("Cancer.x","SYMBOL")) %>% mutate(p.signif=if_else(p.signif=="ns","",p.signif)) %>% spread(Group,MeanValue) %>% mutate(Enrich=if_else(p.signif == "","n.s.",if_else(High>Low,"High","Low"))) %>% gather(Group,Score,High:Low) %>% arrange(Cancer.x,SYMBOL) %>% group_by(Cancer.x,SYMBOL) %>% mutate(Enrich2=if_else(Enrich=="n.s.","n.s.",if_else(Enrich==Group,"Higher","Lower"))) %>% mutate(Cancer.x=factor(Cancer.x,levels = sort(unique(.$Cancer.x)))) %>% mutate(p.signif=factor(p.signif,levels = c("*","**","***","****",""))) %>% mutate(SYMBOL = factor(SYMBOL,levels = rev(unique(sort(.$SYMBOL))))) %>% mutate(Xaxis=paste(Cancer.x,Group,sep = ".")) p1 <- ggplot(Exp.Risk3,aes(x=Xaxis,y=SYMBOL))+ geom_point(aes(size=Score,color=p.signif,shape=Enrich2), stroke=1)+ scale_size_continuous("Mean of log2(TPM+1)",range = c(1,4))+ #facet_wrap(~Attribute,ncol = 11)+ labs(x="",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(#legend.position = "top", axis.text = element_text(color = "black"), axis.text.x = element_blank(), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_manual("P value",values = c(brewer.pal(9,"Set1")[1:4],"#F7F7F7"))+ scale_shape_manual("Score",values = c(1,3,4))+ geom_vline(xintercept = seq(0.5,44.5,2),linetype=2) #scale_fill_manual("Risk score",values = brewer.pal(9,"Set1")[c(1:2)]) p2 <- ggplot(Exp.Risk3 %>% data.frame() %>% dplyr::select(Xaxis,Group) %>% unique(), aes(x=Xaxis,y=1,fill=Group))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("Risk score",values = c("#fe5f55","#004d61")) p3 <- ggplot(Exp.Risk3 %>% data.frame() %>% dplyr::select(Xaxis,Cancer.x) %>% unique(), aes(x=Xaxis,y=1,fill=Cancer.x))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("Cell type",values = Cancer.Col) p123 <- p1 %>% aplot::insert_top(p2,height = 0.05) %>% aplot::insert_top(p3,height = 0.05) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.ImmuneGene.FigureData.pdf",height = 3,width = 12) ggsave(p123,filename = "RiskGroup.COX.TCGA.RiskScore.ImmuneGene.FigureData.Legend.pdf",height = 16,width = 10) ggsave(p,filename = "RiskGroup.COX.TCGA.RiskScore.Estimate.All.pdf",height = 4.5,width = 7) ######## Risk score => Pre+Post => NR+R ########### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(survival) library(survminer) LASSO.Feature <- read.csv("Risk.LASSO.COX.csv") BCAA.ID <- read.csv("BCAA.ID.csv") Feature.Count <- LASSO.Feature %>% filter(Count >= 50) %>% group_by(Cancer) %>% summarise(Count=n()) %>% filter(Count >= 2) Feature.Filter <- LASSO.Feature %>% filter(Count >= 50) %>% filter(Cancer %in% Feature.Count$Cancer) %>% merge(BCAA.ID,by.x = "Feature",by.y = "ENSEMBL") Cal_function <- function(dataset,feature,project,timeVector){ return.list <- list() formula <- as.formula(paste0("Surv(OS.time,OS)~", paste(feature,collapse = "+"))) dataset$OS <- as.numeric(as.double(dataset$OS)) dataset$OS.time <- as.numeric(as.double(dataset$OS.time)) multi_variate_cox <- coxph(formula,data=dataset) dataset$total_risk_score <- predict(multi_variate_cox, type="risk", newdata=dataset[,feature]) return.list[[1]] = dataset COX.Summary <- summary(multi_variate_cox) return.list[[2]] = COX.Summary$concordance[1] library(timeROC) with(dataset, ROC_riskscore <<- timeROC(T = OS.time, delta = OS, marker = total_risk_score, cause = 1, weighting = "marginal", times = timeVector,# c(12,36,60), ROC = TRUE, iid = TRUE) ) return.list[[3]] = ROC_riskscore res.cut <- surv_cutpoint(dataset, time = paste("OS",".time",sep = ""), event = "OS", variables = "total_risk_score") print(res.cut) dat <- surv_categorize(res.cut) #%>% data.frame() %>% mutate(Sample=SNHGs.ICI.data$Sample.ID) fit <- survfit(Surv(OS.time, OS) ~ total_risk_score, data = dat) index="OS" diff=survdiff(as.formula(paste("Surv(",index,".time,",index,") ~ ","total_risk_score",sep = "")),data = dat) pValue=1-pchisq(diff$chisq,df=1) return.list[[4]] = pValue library(ggthemes) theme.sur <- ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=15), axis.text.y = element_text(size=15), axis.title = element_text(size=18), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) p<-ggsurvplot(fit,pval =TRUE, data = dat, surv.median.line = "hv", legend.title = "Risk", conf.int.style = "step", xlab = "Time in days", #break.time.by = 500, risk.table = "abs_pct", #risk.table.y.text.col = T, #risk.table.y.text = FALSE, legend.labs = c("High", "Low"), #pval = TRUE, conf.int = TRUE, palette = "Set1", #ggtheme = theme.sur, risk.table.y.text.col = T, risk.table.y.text = F, ggtheme = theme.sur) print(p) #dev.off() library(export) graph2pdf(file=paste("RiskGroup/Validation.OS.",project,".RiskScore.BestCutoff.KM.pdf",sep = ''),height = 5,width = 4) names(return.list) <- c("Dataset","C.index","timeROC","KM.Pvalue") return(return.list) } ###### TIGER => GBM-PRJNA482620 #### BCAA.ID <- read.csv("BCAA.ID.csv") GBM.Feature <- Feature.Filter %>% filter(Cancer == "LGG") Exp <- read.csv("../../TIGER.Database/GBM-PRJNA482620.Response.tsv",sep = "\t") Clinical <- read.csv("../../TIGER.Database/GBM-PRJNA482620.Response (1).tsv",sep = "\t") #BCAA.ID$SYMBOL %in% Exp$GENE_SYMBOL Exp.Clinical <- Exp %>% filter(GENE_SYMBOL %in% GBM.Feature$SYMBOL) %>% remove_rownames() %>% column_to_rownames("GENE_SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(Clinical,by="sample_id") %>% mutate(OS=if_else(vital.status=="Dead",1,0), OS.time=as.numeric(overall.survival..days.)) DATA <- Cal_function(Exp.Clinical,GBM.Feature$SYMBOL,"GBM-PRJNA482620",c(365,1080,1800)) saveRDS(DATA,file = "ICI.riskScore.Validation.GBM-PRJNA482620.Rds") #### Figure #### p1 <- ggplot(DATA$Dataset %>% filter(total_risk_score<=10), aes(x=response,y=total_risk_score,fill=response))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.4,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(method = "wilcox.test",label.x = 1.2,label = "p.format",size=5)+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p1,filename = "ICI.riskScore.Validation.GBM-PRJNA482620.Boxplot.pdf",height = 2.5,width = 2.5) DATA$Dataset %>% mutate(Group=if_else(total_risk_score > median(total_risk_score),"High","Low")) -> dataset2 fisher.test(table(dataset2$Group,dataset2$response)) ###### TIGER => Melanoma-GSE91061 #### SKCM.Feature <- Feature.Filter %>% filter(Cancer == "SKCM") Exp <- read.csv("../../TIGER.Database/Melanoma-GSE91061.Response.tsv",sep = "\t") Clinical <- read.csv("../../TIGER.Database/Melanoma-GSE91061.Response (1).tsv",sep = "\t") Exp.Clinical <- Exp %>% filter(GENE_SYMBOL %in% GBM.Feature$SYMBOL) %>% remove_rownames() %>% column_to_rownames("GENE_SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(Clinical,by="sample_id") %>% mutate(OS=if_else(vital.status=="Dead",1,0), OS.time=as.numeric(overall.survival..days.)) DATA <- Cal_function(Exp.Clinical,SKCM.Feature$SYMBOL,"Melanoma-GSE91061",c(365,1080,1800)) saveRDS(DATA,file = "ICI.riskScore.Validation.Melanoma-GSE91061.Rds") #### Figure #### DATA$Dataset %>% mutate(RESPONSE = if_else(response %in% c("CR","PR"),"R", if_else(response %in% c("PD","SD"),"NR","UNK"))) %>% mutate(Group=if_else(total_risk_score > median(total_risk_score),"High","Low")) -> dataset2 p1 <- ggplot(dataset2 , #%>% filter(Treatment=="ON") %>% filter(total_risk_score<=10) aes(x=RESPONSE,y=total_risk_score,fill=RESPONSE))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.2,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(comparisons = list(c("NR","R")),label.x = 1,label = "p.format",size=5, method="wilcox.test")+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p1,filename = "ICI.riskScore.Validation.Melanoma-GSE91061.Boxplot.pdf",height = 2.5,width = 2.5) fisher.test(table(dataset2$Group,dataset2$RESPONSE)) p2 <- ggplot(dataset2 , #%>% filter(Treatment=="ON") %>% filter(total_risk_score<=10) aes(x=Treatment,y=total_risk_score,fill=Treatment))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.2,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(label.x = 1,label = "p.format",size=5, method="wilcox.test")+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p2,filename = "ICI.riskScore.Validation.Melanoma-GSE91061.Boxplot.PREON.pdf",height = 2.5,width = 2.5) ###### TIGER => Melanoma-phs000452 #### SKCM.Feature <- Feature.Filter %>% filter(Cancer == "SKCM") Exp <- read.csv("../../TIGER.Database/Melanoma-phs000452.Response.tsv",sep = "\t") Clinical <- read.csv("../../TIGER.Database/Melanoma-phs000452.Response (1).tsv",sep = "\t") Exp.Clinical <- Exp %>% filter(GENE_SYMBOL %in% SKCM.Feature$SYMBOL) %>% remove_rownames() %>% column_to_rownames("GENE_SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(Clinical,by="sample_id") %>% mutate(OS=if_else(vital.status=="Dead",1,0), OS.time=as.numeric(overall.survival..days.)) DATA <- Cal_function(Exp.Clinical,SKCM.Feature$SYMBOL,"Melanoma-phs000452",c(365,1080,1800)) saveRDS(DATA,file = "ICI.riskScore.Validation.Melanoma-phs000452.Rds") #### Figure #### DATA$Dataset %>% #mutate(RESPONSE = if_else(response %in% c("CR","PR"),"R",if_else(response %in% c("PD","SD"),"NR","UNK"))) %>% mutate(Group=if_else(total_risk_score > median(total_risk_score),"High","Low")) -> dataset2 p1 <- ggplot(dataset2 %>% filter(Treatment=="EDT"), # %>% filter(total_risk_score<=10) aes(x=response_NR,y=total_risk_score,fill=response_NR))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.2,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(comparisons = list(c("N","R")),label.x = 1,label = "p.format",size=5, method="wilcox.test")+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p1,filename = "ICI.riskScore.Validation.Melanoma-phs000452.Boxplot.pdf",height = 2.5,width = 2.5) fisher.test(table(dataset2$Group,dataset2$response_NR)) ###### TIGER => Melanoma-PRJEB23709 #### SKCM.Feature <- Feature.Filter %>% filter(Cancer == "SKCM") Exp <- read.csv("../../TIGER.Database/Melanoma-PRJEB23709.Response.tsv",sep = "\t") Clinical <- read.csv("../../TIGER.Database/Melanoma-PRJEB23709.Response (1).tsv",sep = "\t") Exp.Clinical <- Exp %>% filter(GENE_SYMBOL %in% SKCM.Feature$SYMBOL) %>% remove_rownames() %>% column_to_rownames("GENE_SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(Clinical,by="sample_id") %>% mutate(OS=if_else(vital.status=="Dead",1,0), OS.time=as.numeric(overall.survival..days.)) DATA <- Cal_function(Exp.Clinical,SKCM.Feature$SYMBOL,"Melanoma-PRJEB23709",c(365,1080,1800)) saveRDS(DATA,file = "ICI.riskScore.Validation.Melanoma-PRJEB23709.Rds") #### Figure #### DATA$Dataset %>% #mutate(RESPONSE = if_else(response %in% c("CR","PR"),"R",if_else(response %in% c("PD","SD"),"NR","UNK"))) %>% mutate(Group=if_else(total_risk_score > median(total_risk_score),"High","Low")) -> dataset2 p1 <- ggplot(dataset2 , #%>% filter(Treatment=="ON") %>% filter(total_risk_score<=10) aes(x=response_NR,y=total_risk_score,fill=response_NR))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.2,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(comparisons = list(c("N","R")),label.x = 1,label = "p.format",size=5, method="wilcox.test")+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p1,filename = "ICI.riskScore.Validation.Melanoma-phs000452.Boxplot.pdf",height = 2.5,width = 2.5) fisher.test(table(dataset2$Group,dataset2$response_NR)) ###### TIGER => RCC-Braun_2020 ##### KIRC.Feature <- Feature.Filter %>% filter(Cancer == "KIRC") Exp <- read.csv("../../TIGER.Database/RCC-Braun_2020.Response.tsv",sep = "\t") Clinical <- read.csv("../../TIGER.Database/RCC-Braun_2020.Response (1).tsv",sep = "\t") Exp.Clinical <- Exp %>% filter(GENE_SYMBOL %in% KIRC.Feature$SYMBOL) %>% remove_rownames() %>% column_to_rownames("GENE_SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(Clinical,by="sample_id") %>% mutate(OS=if_else(vital.status=="Dead",1,0), OS.time=as.numeric(overall.survival..days.)) DATA <- Cal_function(Exp.Clinical,KIRC.Feature$SYMBOL,"RCC-Braun_2020",c(365,1080,1800)) saveRDS(DATA,file = "ICI.riskScore.Validation.RCC-Braun_2020.Rds") DATA$Dataset %>% #mutate(RESPONSE = if_else(response %in% c("CR","PR"),"R",if_else(response %in% c("PD","SD"),"NR","UNK"))) %>% mutate(Group=if_else(total_risk_score > median(total_risk_score),"High","Low")) -> dataset2 p1 <- ggplot(dataset2 , #%>% filter(Treatment=="ON") %>% filter(total_risk_score<=10) aes(x=response_NR,y=total_risk_score,fill=response_NR))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.2,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(comparisons = list(c("N","R")),label.x = 1,label = "p.format",size=5, method="wilcox.test")+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p1,filename = "ICI.riskScore.Validation.RCC-Braun_2020.Boxplot.pdf",height = 2.5,width = 2.5) ###### IMvigor210 ##### BLCA.Feature <- Feature.Filter %>% filter(Cancer == "BLCA") load("../../IMvigor210CoreBiologies/IMvigor210CoreBiologies.Rdata") # annoData是基因的注释信息 # phenoData是样本信息 # expreSet 是表达矩阵，行是基因31286，列是样本348 SYMBOLS <- annoData %>% filter(symbol %in% KIRC.Feature$SYMBOL) %>% dplyr::select(entrez_id,symbol) Exp.Clinical <- expreSet[SYMBOLS$entrez_id,] %>% data.frame(check.names = F) %>% rownames_to_column("entrez_id") %>% merge(SYMBOLS,by="entrez_id") %>% dplyr::select(-entrez_id) %>% remove_rownames() %>% column_to_rownames("symbol") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("sample_id") %>% merge(phenoData %>% rownames_to_column("sample_id"),by="sample_id") %>% mutate(OS=censOS, OS.time=as.numeric(os)) DATA <- Cal_function(Exp.Clinical,BLCA.Feature$SYMBOL,"IMvigor210",c(1,3,5)) saveRDS(DATA,file = "ICI.riskScore.Validation.IMvigor210.Rds") DATA$Dataset %>% #mutate(RESPONSE = if_else(response %in% c("CR","PR"),"R",if_else(response %in% c("PD","SD"),"NR","UNK"))) %>% mutate(Group=if_else(total_risk_score > 0.9881938,"High","Low")) %>% mutate(RESPONSE=if_else(`Best Confirmed Overall Response` %in% c("CR","PR"),"R", if_else(`Best Confirmed Overall Response` %in% c("PD","SD"),"NR","UNK"))) -> dataset2 p1 <- ggplot(dataset2 %>% filter(RESPONSE != "UNK") , #%>% filter(Treatment=="ON") %>% filter(total_risk_score<=10) aes(x=RESPONSE,y=total_risk_score,fill=RESPONSE))+ geom_violin(cex=1.3)+ geom_boxplot(width=0.2,color="white")+ labs(x="",y="Risk score")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=12), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ stat_compare_means(comparisons = list(c("NR","R")),label.x = 1,label = "p.format",size=5, method="wilcox.test")+ #scale_x_discrete(labels=c("NR","R"))+ ggsci::scale_fill_npg() ggsave(p1,filename = "ICI.riskScore.Validation.IMvigor210.Boxplot.pdf",height = 2.5,width = 2.5) dataset3 <- dataset2 %>% filter(RESPONSE != "UNK") fisher.test(table(dataset3$Group,dataset3$RESPONSE)) fisher.test(matrix(c(80,20,76,24),ncol = 2)) ######## Drugs => druggable gene mRNA => 46 genes >= 10 Cancers ########### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) Drug.Gene <- data.table::fread("../../DrugsDGIDB/genes.tsv") GeneType <- read.table("../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,HGNC.symbol,Gene.type,Gene.name,NCBI.gene..formerly.Entrezgene..ID) %>% unique() GeneType.Unique %>% filter(Gene.type == "protein_coding") %>% dplyr::select(Gene.stable.ID,HGNC.symbol) %>% filter(HGNC.symbol %in% Drug.Gene$gene_name) %>% na.omit() %>% mutate(HGNC.symbol=as.character(HGNC.symbol)) %>% filter(HGNC.symbol != "") %>% magrittr::set_colnames(c("ENSEMBL","SYMBOL")) -> druggene RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") dirs <- list.dirs("../../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] for (dir in dirs) { Exp.Spearman <- data.frame() project <- str_remove(dir,".*/") if (project %in% RiskScore.Cancer$Cancer) { print(project) phenotype <- read.csv(paste(dir,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) Exp.Risk <- TPM.log[intersect(rownames(TPM.log),druggene$ENSEMBL) %>% sort()%>%unique(),] %>% data.frame(check.names = F) %>% #rownames_to_column("ENSEMBL") %>% #merge(druggene,by="ENSEMBL") %>% #dplyr::select(-ENSEMBL) %>% #column_to_rownames("SYMBOL") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("Sample.ID") %>% filter(Sample.ID %in% PhenoType$Sample.ID) %>% #mutate(Cancer=project) %>% merge(RiskScore.Cancer,by.x="Sample.ID",by.y="SampleID") for (gene in intersect(rownames(TPM.log),druggene$ENSEMBL) %>% sort()%>%unique()) { print(gene) test <- cor.test(Exp.Risk[[gene]],Exp.Risk$total_risk_score, method = "spearman",exact = F) Exp.Spearman <- data.frame(ENSEMBL=gene, Rho=test$estimate, Pvalue=test$p.value, Cancer=project) %>% rbind.data.frame(Exp.Spearman) } } write.csv(Exp.Spearman,file = paste("Drugs.DGI.",project,".csv",sep = ""),row.names = F) } #Exp.Spearman %>% filter(Cancer!="LIHC") %>% write.csv("Drugs.DGI.Part1.csv",row.names = F) library(tidyverse) Exp.Spearman <- data.frame() files <- list.files(".",pattern = "^Drugs") files <- files[str_detect(files,"DGI")] for (file in files) { data <- read.csv(file) Exp.Spearman <- rbind.data.frame(Exp.Spearman,data) } write.csv(Exp.Spearman,"DRUGs.DGI.All.csv",row.names = F) #### Figure #### Exp.Spearman <- read.csv("DRUGs.DGI.All.csv") n=length(unique(Exp.Spearman$Cancer)) dim(Exp.Spearman) Selec.Gene <- Exp.Spearman %>% filter(Pvalue <= 0.05 & abs(Rho) >= 0.3) %>% mutate(Type=if_else(Rho>0,"P","N")) %>% filter(! ENSEMBL %in% BCAA.ID$ENSEMBL) %>% #filter(Type=="P") %>% group_by(ENSEMBL,Type) %>% summarise(Count=n()) %>% arrange(desc(Count)) %>% filter(Count >= 10) %>% #top_n(100,Count) %>% arrange(Count) %>% merge(druggene,by="ENSEMBL") write.csv(Selec.Gene,"CERES.SelectGene.csv",row.names = F) p1 <- Selec.Gene %>% #top_n(7,Count) %>% filter(SYMBOL %in% CERES.Score1$SYMBOL) %>% ggplot(aes(x=Count,y=SYMBOL,fill=SYMBOL))+ geom_col()+ labs(x="No. of cancers",y="")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggthemes::scale_fill_stata()+ geom_text(aes(label=Count))+ coord_flip() ggsave(p1,filename = "DRUGs.DGI.Positive.pdf",width = 3,height = 2) ######## CRISPR dependency score (CERES) ############ CERES.data <- read.csv("../../DepMap/CRISPR_(Project_Score,_CERES).csv") CERES.Target <- CERES.data %>% dplyr::select(intersect(Selec.Gene$SYMBOL,colnames(CERES.data))) CERES.Score1 <- CERES.Target %>% gather(SYMBOL,Score) %>% filter(SYMBOL %in% Selec.Gene$SYMBOL) %>% group_by(SYMBOL) %>% summarise(Score.M = median(Score)) %>% filter(Score.M <= -1) CERES.Score2 <- CERES.Target %>% gather(SYMBOL,Score) %>% filter(SYMBOL %in% Selec.Gene$SYMBOL) %>% filter(SYMBOL %in% CERES.Score1$SYMBOL) p2 <- ggplot(CERES.Score2,aes(x=reorder(SYMBOL,Score),y=Score,fill=SYMBOL))+ geom_boxplot()+ #geom_col()+ labs(x="",y="CERES")+ #facet_wrap(~Attribute,scale="free_x",ncol = 15)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggthemes::scale_fill_stata()+ geom_hline(yintercept = c(-1),linetype=2) ggsave(p2,filename = "DRUGs.DGI.Positive.13.SYMBOL.CERES.pdf",width = 4,height = 2) ######## RISK SCORE => DEGs ######### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% group_by(Cancer) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group=if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") dir.create("DEGs.RiskScore") files <- list.files("../../mRNA.Pancancer.Exp.Count/",pattern = "csv$") #files <- list.files("../mRNA.Pancancer.Exp.Count/",pattern = "csv$") for (file in files) { project <- file %>%str_remove("PanCancer\\.") %>% str_remove("\\..*") %>% str_remove("TCGA-") print(project) if (project %in% RiskScore.Cancer$Cancer) { #Count.data <- read.csv(file = file.path("../../mRNA.Pancancer.Exp.Count/",file),check.names = F,row.names = 1) Count.data <- read.csv(file = file.path("../../mRNA.Pancancer.Exp.Count/",file),check.names = F,row.names = 1) #filename <- paste("TCGA-",project,".mRNA.TPM.csv",sep = "") #TPM.data <- read.csv(file = file.path("../../mRNA.PanCancer.Exp/",filename),check.names = F,row.names = 1) #TPM.data <- read.csv(file = file.path("../../mRNA.PanCancer.Exp/",filename),check.names = F,row.names = 1) Count.data <- Count.data %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("SampleID") %>% filter(SampleID %in% RiskScore.Cancer$SampleID) %>% column_to_rownames("SampleID") %>% t() %>% data.frame(check.names = F) countData <- Count.data[rowMeans(Count.data)>1,] colData <- data.frame(SampleID=colnames(Count.data)) %>% merge(RiskScore.Cancer %>% dplyr::select(SampleID,Group),by="SampleID") %>% column_to_rownames("SampleID") colData$Group <- factor(colData$Group,levels = c("High","Low")) countData <- countData %>% dplyr::select(rownames(colData)) #condition <- factor(GENE.EXP$Group,levels = c("high","low")) #colData <- data.frame(row.names=colnames(countData), condition) library(DESeq2) dds <- DESeqDataSetFromMatrix(countData = countData, colData = colData, design = ~ Group) dds1 <- DESeq(dds) res <- results(dds1,contrast=c("Group","High","Low")) res1 <- data.frame(res, stringsAsFactors = FALSE, check.names = FALSE) outfilename <- paste(project,".DEG.RiskScore.csv",sep = "") write.csv(res1,file = file.path("DEGs.RiskScore/",outfilename)) } } #### Dataset2 #### Exp.Data <- data.frame() files <- list.files("DEGs.RiskScore/",pattern = "csv$") for (file in files) { project <- file %>% str_remove_all("\\..*") print(project) Exp.Data <- read.csv(file.path("DEGs.RiskScore/",file)) %>% mutate(Cancer=project) %>% rbind.data.frame(Exp.Data) } GeneType <- read.table("../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,HGNC.symbol,Gene.type,Gene.name,NCBI.gene..formerly.Entrezgene..ID) %>% unique() GeneType.Unique %>% #filter(Gene.type == "protein_coding") %>% dplyr::select(Gene.stable.ID,HGNC.symbol) %>% filter(HGNC.symbol %in% Drug.Gene$gene_name) %>% na.omit() %>% mutate(HGNC.symbol=as.character(HGNC.symbol)) %>% filter(HGNC.symbol != "") %>% magrittr::set_colnames(c("ENSEMBL","SYMBOL")) -> druggene Up.Gene <- Exp.Data %>% filter(padj <= 0.05) %>% filter(log2FoldChange >= 1) %>% #mutate(Type=if_else(log2FoldChange >= 1,"Up","Down")) %>% group_by(X) %>% summarise(Count=n()) %>% arrange(desc(Count)) %>% filter(Count >= 5) %>% merge(druggene,by.x="X",by.y="ENSEMBL") %>% top_n(100,Count) Down.Gene <- Exp.Data %>% filter(padj <= 0.05) %>% filter(log2FoldChange <= -1) %>% #mutate(Type=if_else(log2FoldChange >= 1,"Up","Down")) %>% group_by(X) %>% summarise(Count=n()) %>% arrange(desc(Count)) %>% filter(Count >= 5) %>% merge(druggene,by.x="X",by.y="ENSEMBL") write.csv(Up.Gene %>% dplyr::select(SYMBOL),"CMAP.UP.SYMBOL2.csv",row.names = F) write.csv(Down.Gene %>% dplyr::select(SYMBOL),"CMAP.DOWN.SYMBOL2.csv",row.names = F) GPL96=read.table("GPL96-57554.txt", header = TRUE,fill = T,sep = "\t", comment.char = "#", stringsAsFactors = FALSE, quote = "") Up.Gene.Trans <- GPL96[GPL96$Gene.Symbol %in% Up.Gene$SYMBOL,c(1,11)] write.csv(Up.Gene.Trans,"CMAP.UP.SYMBOL.csv",row.names = F) Down.Gene.Trans <- GPL96[GPL96$Gene.Symbol %in% Down.Gene$SYMBOL,c(1,11)] write.csv(Down.Gene.Trans,"CMAP.DOWN.SYMBOL.csv",row.names = F) #### Confirm SYMBOLS => SPIED3 + CMAP #### GeneType <- read.table("../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,HGNC.symbol,Gene.type,Gene.name,NCBI.gene..formerly.Entrezgene..ID) %>% unique() GeneType.Unique %>% filter(Gene.type == "protein_coding") %>% dplyr::select(Gene.stable.ID,HGNC.symbol) %>% #filter(HGNC.symbol %in% Drug.Gene$gene_name) %>% na.omit() %>% mutate(HGNC.symbol=as.character(HGNC.symbol)) %>% filter(HGNC.symbol != "") %>% magrittr::set_colnames(c("ENSEMBL","SYMBOL")) -> druggene files <- list.files("DEGs.RiskScore/",pattern = "csv$") files <- files[str_detect(files,"DEG")] for (file in files) { project <- file %>% str_remove_all("\\..*") print(project) data1 <- read.csv(file.path("DEGs.RiskScore/",file)) %>% mutate(Cancer=project) %>% filter(padj <= 0.05) %>% filter(abs(log2FoldChange) >= 1) %>% merge(druggene,by.x="X",by.y="ENSEMBL") %>% dplyr::select(SYMBOL,log2FoldChange) %>% arrange(desc(log2FoldChange)) %>% filter() data2 <- read.csv(file.path("DEGs.RiskScore/",file)) %>% mutate(Cancer=project) %>% filter(padj <= 0.05) %>% filter(abs(log2FoldChange) >= 1) %>% merge(druggene,by.x="X",by.y="ENSEMBL") %>% dplyr::select(SYMBOL,log2FoldChange) %>% arrange(desc(log2FoldChange)) %>% filter() write.table(file = file.path("DEGs.RiskScore/", paste(project,".Sig.CMAP.txt",sep = "")), row.names = F,col.names = F,sep = "\t",quote = F) } #### CMAP => DRUGS #### setwd("K:/TCGA/Anlysis/BCAA/CMAP") library(data.table) library(tidyverse) files <- list.files("./",recursive = T) files <- files[str_detect(files,"query_result")] CMAP.GRUGS <- data.frame() for (file in files) { project = file %>% str_remove("\\..*") print(project) data <- read.table(file, skip = 2,sep = "\t",header=T) data <- data %>% filter(pert_type == "trt_cp") %>% mutate(fdr_q_nlog10=as.numeric(fdr_q_nlog10)) %>% arrange(desc(fdr_q_nlog10)) %>% filter(fdr_q_nlog10 >= 10) CMAP.GRUGS <- rbind.data.frame(CMAP.GRUGS,data%>%mutate(Cancer=project)) write.csv(data,file = paste("CMAP.",project,".csv",sep = ""),row.names = F) } Anno <- fread("CMAP.Drugs.Repurposing_Hub_export.txt") Class <- CMAP.GRUGS %>% #merge(Anno,by.x="pert_iname",by.y="Name",all.x = T) %>% group_by(moa) %>% summarise(Count=n()) %>% arrange(desc(Count)) %>% filter(moa != -666) %>% top_n(10,Count) write.csv(CMAP.GRUGS,file = "CMAP.Drugs.csv",row.names = F) p1 <- ggplot(Class,aes(x=Count,y=reorder(moa,Count),fill=moa))+ geom_col()+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggthemes::scale_fill_stata()+ geom_text(aes(label=Count),hjust=1)+ labs(y="",x="Frequence") ggsave(p1,filename = "CMAP.pdf",height = 3,width = 3) # c("Fluorouracil", "Cisplatin", "Oxaliplatin", "Sorafenib","Docetaxel", "Irinotecan", "Capecitabine") ###### oncoPredict ############## setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(oncoPredict) library(data.table) library(gtools) library(reshape2) library(ggpubr) library(ggplot2) dir.create("oncoPredict") options("expressions"=20000) #memory.limit(size=8000000) dir1 = "../../oncoPredict/DataFiles/DataFiles/Training Data/" GDSC2_Expr = readRDS(file=file.path(dir1,'GDSC2_Expr (RMA Normalized and Log Transformed).rds')) GDSC2_Res = readRDS(file = file.path(dir1,"GDSC2_Res.rds")) GDSC2_Res <- exp(GDSC2_Res) CTRP2_Expr = readRDS(file=file.path(dir1,'CTRP2_Expr (TPM, not log transformed).rds')) CTRP2_Res = readRDS(file = file.path(dir1,"CTRP2_Res.rds")) CTRP2_Res <- exp(CTRP2_Res) #testExpr<- GDSC2_Expr[,sample(1:ncol(GDSC2_Expr),10)] #colnames(testExpr)=paste0('test',colnames(testExpr)) #testExpr[1:4,1:4] RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") dirs <- list.dirs("../../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] for (dir in dirs) { project <- str_remove(dir,".*/") print(project) if (project %in% RiskScore.Cancer$Cancer) { #phenotype <- read.csv(paste(dir,"/TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% # dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) #TPM.log <- log2(TPM+1) G_list <- read.csv("../LIHC/biomaRt.GeneID.tansfer.csv") mRNA.Exp.Target <- TPM %>% #[,intersect(phenotype$submitter_id.samples,colnames(TPM.log))] %>% data.frame(check.names = F) t() %>% data.frame(check.names = F) %>% rownames_to_column("SampleID") %>% filter(SampleID %in% PhenoType$Sample.ID) %>% filter(SampleID %in% RiskScore.Cancer$SampleID) %>% remove_rownames() %>% column_to_rownames("SampleID") %>% t() %>% data.frame(check.names = F) dat.2 <- mRNA.Exp.Target %>% data.frame(check.names = F) %>% rownames_to_column("ensembl_gene_id") %>% data.frame(check.names = F) %>% merge(.,G_list[,1:2],by="ensembl_gene_id") %>% data.frame(check.names = F) %>% na.omit() %>% dplyr::select(-ensembl_gene_id) expr_mean=aggregate(.~hgnc_symbol,mean,data=dat.2) #calculate mean for same symbol dat.3 <- expr_mean %>% remove_rownames() %>% column_to_rownames("hgnc_symbol") if(! file.exists(paste("oncoPredict/GDSC2.",project,".drugSensitivity.csv",sep = ""))){ calcPhenotype(trainingExprData = GDSC2_Expr, trainingPtype = GDSC2_Res, testExprData = as.matrix(dat.3), batchCorrect = 'eb', # "eb" for ComBat powerTransformPhenotype = TRUE, removeLowVaryingGenes = 0.2, minNumSamples = 10, printOutput = TRUE, removeLowVaringGenesFrom = 'homogenizeData' ) testPtype <- fread('./calcPhenotype_Output/DrugPredictions.csv', data.table = F) write.csv(testPtype,file = paste("oncoPredict/GDSC2.",project,".drugSensitivity.csv",sep = ""),quote = F) } if(! file.exists(paste("oncoPredict/CTRP2.",project,".drugSensitivity.csv",sep = ""))){ calcPhenotype(trainingExprData = CTRP2_Expr, trainingPtype = CTRP2_Res, testExprData = as.matrix(dat.3), batchCorrect = 'eb', # "eb" for ComBat powerTransformPhenotype = TRUE, removeLowVaryingGenes = 0.2, minNumSamples = 10, printOutput = TRUE, removeLowVaringGenesFrom = 'rawData' ) testPtype <- fread('./calcPhenotype_Output/DrugPredictions.csv', data.table = F) write.csv(testPtype,file = paste("oncoPredict/CTRP2.",project,".drugSensitivity.csv",sep = ""),quote = F) } print("DONE") } } calcPhenotype(trainingExprData = GDSC2_Expr, trainingPtype = GDSC2_Res, testExprData = as.matrix(dat.3), batchCorrect = 'eb', # "eb" for ComBat powerTransformPhenotype = TRUE, removeLowVaryingGenes = 0.2, minNumSamples = 10, printOutput = TRUE, removeLowVaringGenesFrom = 'rawData' ) library(data.table) testPtype <- fread('./calcPhenotype_Output/DrugPredictions.csv', data.table = F) ###### GDSC2 ###### setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% group_by(Cancer) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group=if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() files <- list.files("oncoPredict/GDSC/",pattern = "csv$") Drugs.Wilcox <- data.frame() for (file in files) { project = (file %>% str_split("\\.") %>% unlist())[2] if (project %in% RiskScore.Cancer$Cancer) { print(project) library(rstatix) data <- read.csv(file.path("oncoPredict/GDSC/",file),row.names = 1) %>% merge(RiskScore.Cancer,by.x = "V1",by.y="SampleID") %>% gather(Drugs,Sensitivity,Camptothecin_1003:JQ1_2172) data2 <- data %>% na.omit() %>% group_by(Group,Drugs) %>% summarise(S.M = median(Sensitivity)) %>% ungroup() %>% spread(Group,S.M) library(ggpubr) Drugs.Wilcox <- compare_means(Sensitivity~Group,data = data,group.by = "Drugs",method = "wilcox") %>% mutate(Cancer=project) %>% merge(data2,by="Drugs") %>% rbind.data.frame(Drugs.Wilcox) } } write.csv(Drugs.Wilcox,file = "Pancancer.GDSC2.Drugs.Sensitivity.csv",row.names = F) #### Figure #### Drugs.Wilcox2 <- Drugs.Wilcox %>% filter(p.signif != "ns") %>% mutate(Label = if_else(High >= Low,"High","Low")) %>% mutate(Drugs2 = str_remove_all(Drugs,"_.*")) %>% #filter(Drugs2 == "Docetaxel") group_by(Drugs,Label) %>% summarise(Count=n()) %>% ungroup() %>% arrange(desc(Count)) %>% #filter(Drugs == "Docetaxel") filter(Count >= 10) #Phenotype <- data.frame(Drugs.Wilcox2$Drugs,Rename=c("BRD Inhibitor","PDGFR tyrosine kinase receptor inhibitor", # "K-Ras(G12C)-inhibitor-12", # "Protein Kinase inhibitor", # "Mitochondrial inhibitor", # "ATM Kinase Inhibitor", # "Protein tyrosine kinases inhibitor", # "Unknown", # "RAF inhibitor", # )) p1 <- ggplot(data=Drugs.Wilcox %>% mutate(Label = if_else(p.signif=="ns","N.S.", if_else(High >= Low,"High","Low"))) %>% filter(Drugs %in% Drugs.Wilcox2$Drugs), aes(x=Cancer,y=Drugs,color=Label,shape=Label) )+ geom_point(size=2)+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_npg()+ scale_color_manual(values = c("#E64B35FF","#3C5488FF","#B09C85FF"))+ scale_shape_manual(values = c(1,3,4))+ labs(x="",y="",color="Drug sensitivity\nfor risk group", shape="Drug sensitivity\nfor risk group") p2 <- ggplot(data=Drugs.Wilcox %>% mutate(Label = if_else(p.signif=="ns","N.S.", if_else(High >= Low,"High","Low"))) %>% filter(Drugs %in% Drugs.Wilcox2$Drugs) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% mutate(Label=factor(Label,levels = rev(c("High","Low","N.S.")))), aes(x=Count,y=Drugs,fill=Label))+ geom_col()+ scale_fill_manual(values = rev(c("#E64B35FF","#3C5488FF","#B09C85FF")))+ labs(x="No. of cancer",y="")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) p<-p1 %>% aplot::insert_right(p2,width = 0.3) ggsave(p,filename = "Drugs.GDSC2.Grugs.Sensitivity.pdf",height = 5,width = 7) ###### PRISM + CTRP => OncoPredict ########## setwd("K:/TCGA/Anlysis/BCAA") library(tidyverse) library(oncoPredict) library(data.table) library(gtools) library(reshape2) library(ggpubr) library(ggplot2) #library(impute) #library(SimDesign) # Information used to suppress the output of the susceptibility prediction process prism.auc <- read.csv("DepMap/Drug_sensitivity_AUC_(PRISM_Repurposing_Secondary_Screen).csv", check.names = F,stringsAsFactors = F,row.names = 1) ctrp.auc <- read.csv("DepMap/Drug_sensitivity_AUC_(CTD^2).csv", check.names = F,stringsAsFactors = F,row.names = 1) #celline.anno <- read.csv("DepMap/CBX2 Expression Public 22Q4.csv") ## Remove drugs with more than 20% missing values prism.auc <- prism.auc[,apply(prism.auc,2,function(x) sum(is.na(x))) < 0.2*nrow(prism.auc)] ctrp.auc <- ctrp.auc[,apply(ctrp.auc,2,function(x) sum(is.na(x))) < 0.2*nrow(ctrp.auc)] ### Load expression profiles of CCLE cell lines as training set ccle.exp <- read.csv("DepMap/DepMap.22Q2.TPM.Processed.csv", check.names = F,row.names = 1) PhenoType <- read.csv("../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") dirs <- list.dirs("../Clinical.XENA/",full.names = T) dirs <- dirs[str_detect(dirs,"//")] for (dir in dirs) { project <- str_remove(dir,".*/") print(project) TPM <- read.csv(paste("../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) #TPM.log <- log2(TPM+1) G_list <- read.csv("biomaRt.GeneID.tansfer.csv") mRNA.Exp.Target <- TPM %>% #[,intersect(phenotype$submitter_id.samples,colnames(TPM.log))] %>% data.frame(check.names = F) t() %>% data.frame(check.names = F) %>% rownames_to_column("SampleID") %>% filter(SampleID %in% PhenoType$Sample.ID) %>% #filter(SampleID %in% RiskScore.Cancer$SampleID) %>% remove_rownames() %>% column_to_rownames("SampleID") %>% t() %>% data.frame(check.names = F) dat.2 <- mRNA.Exp.Target %>% data.frame(check.names = F) %>% rownames_to_column("ensembl_gene_id") %>% data.frame(check.names = F) %>% merge(.,G_list[,1:2],by="ensembl_gene_id") %>% data.frame(check.names = F) %>% na.omit() %>% dplyr::select(-ensembl_gene_id) expr_mean=aggregate(.~hgnc_symbol,mean,data=dat.2) #calculate mean for same symbol dat.3 <- expr_mean %>% remove_rownames() %>% column_to_rownames("hgnc_symbol") keepgene <- apply(dat.3, 1, mad) > 0.5 testExpr <- log2(dat.3[keepgene,] + 1) comgene <- intersect(rownames(ccle.exp),rownames(testExpr)) trainExpr <- as.matrix(ccle.exp[comgene,]) testExpr <- testExpr[comgene,] if (! file.exists(paste("OncoPredict/CTRP.AUC.",project,".csv",sep = ""))) { trainPtype <- ctrp.auc ### The loop is slow, please be patient calcPhenotype(trainingExprData = as.matrix(trainExpr), trainingPtype = as.matrix(trainPtype), testExprData = as.matrix(testExpr), ##"testExprData" must be a matrix. powerTransformPhenotype = F, selection = 2,batchCorrect = "eb", printOutput = T,removeLowVaringGenesFrom = "rawData", minNumSamples = 10) ctrp.pred.auc <- fread('./calcPhenotype_Output/DrugPredictions.csv', data.table = F) write.csv(ctrp.pred.auc,file=paste("OncoPredict/CTRP.AUC.",project,".csv",sep = ""),quote=F, col.names=T, row.names=T) } if (! file.exists(paste("OncoPredict/PRISM.AUC.",project,".csv",sep = ""))) { trainPtype <- prism.auc ### The loop is slow, please be patient calcPhenotype(trainingExprData = as.matrix(trainExpr), trainingPtype = as.matrix(trainPtype), testExprData = as.matrix(testExpr), ##"testExprData" must be a matrix. powerTransformPhenotype = F, removeLowVaringGenesFrom = "rawData", printOutput = T, selection = 2,batchCorrect = "eb", minNumSamples = 10) prism.pred.auc <- fread('./calcPhenotype_Output/DrugPredictions.csv', data.table = F) write.csv(prism.pred.auc,file=paste("OncoPredict/PRISM.AUC.",project,".csv",sep = ""),quote=F, col.names=T, row.names=T) } } calcPhenotype(trainingExprData = trainExpr, trainingPtype = CTRP2_Res, testExprData = as.matrix(testExpr), batchCorrect = 'eb', # "eb" for ComBat powerTransformPhenotype = F, removeLowVaryingGenes = 0.2, minNumSamples = 10, printOutput = TRUE) ###### PRISM ####### setwd("K:/TCGA/Anlysis/BCAA/Depmap") files <- list.files(".",pattern = "csv$") files <- files[str_detect(files,"PRISM")] RiskScore.Cancer <- read.csv(file = "../RiskGroup.COX.TCGA.RiskScore.csv") %>% group_by(Cancer) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group=if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() PRISM.RiskScore <- data.frame() for (file in files) { project = (file %>% str_split("\\.") %>% unlist())[3] print(project) if (project %in% RiskScore.Cancer$Cancer) { data <- read.csv(file,row.names = 1,check.names = F) %>% magrittr::set_colnames(str_remove_all(colnames(.),"\\(.*")) %>% merge(RiskScore.Cancer,by.x="V1",by.y="SampleID") High.G <- data %>% filter(Group == "High") Low.G <- data %>% filter(Group == "Low") for (drug in colnames(data)[2:432]) { if (length(na.omit(data[[drug]])) > 0) { test <- wilcox.test(High.G[[drug]],Low.G[[drug]],alternative = "two.sided", exact = F) PRISM.RiskScore <- data.frame(Pvalue=test$p.value, High.Mean=mean(High.G[[drug]]), High.Median=median(High.G[[drug]]), Low.Mean=mean(Low.G[[drug]]), Low.Median=median(Low.G[[drug]]), Drugs = drug, Cancer=project) %>% rbind.data.frame(PRISM.RiskScore) } } } } write.csv(PRISM.RiskScore,"Drugs.Depmap.PRISM.Wilcox.csv",row.names = F) PRISM.RiskScore.S <- PRISM.RiskScore %>% filter(Pvalue <= 0.05) %>% mutate(Label = if_else(High.Median>Low.Median,"High","Low")) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% arrange(desc(Count))%>% mutate(Drugs=str_remove_all(Drugs," ")) #### Figure #### PRISM.RiskScore.S <- PRISM.RiskScore %>% #filter(Pvalue <= 0.05) %>% mutate(Label = if_else(Pvalue>0.05,"N.S.", if_else(High.Median>Low.Median,"High","Low"))) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% arrange(desc(Count))%>% filter(Label!="N.S.") %>% mutate(Drugs=str_remove_all(Drugs," ")) %>% filter(Count > 12) PRISM.RiskScore2 <- PRISM.RiskScore %>% mutate(Drugs=str_remove_all(Drugs," ")) %>% mutate(Label = if_else(Pvalue>0.05,"N.S.", if_else(High.Median>Low.Median,"High","Low"))) %>% filter(Drugs %in% PRISM.RiskScore.S$Drugs) PRISM.RiskScore3 <- PRISM.RiskScore %>% #filter(Pvalue <= 0.05) %>% mutate(Label = if_else(Pvalue>0.05,"N.S.", if_else(High.Median>Low.Median,"High","Low"))) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% arrange(desc(Count))%>% #filter(Label!="N.S.") %>% mutate(Drugs=str_remove_all(Drugs," ")) %>% filter(Drugs %in% PRISM.RiskScore.S$Drugs) p1 <- ggplot(data=PRISM.RiskScore2, aes(x=Cancer,y=Drugs,color=Label,shape=Label) )+ geom_point(size=2)+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_npg()+ scale_color_manual(values = c("#E64B35FF","#3C5488FF","#B09C85FF"))+ scale_shape_manual(values = c(1,3,4))+ labs(x="",y="",color="Predicted AUC\nfor risk group", shape="Predicted AUC\nfor risk group") p2 <- ggplot(data=PRISM.RiskScore3 %>% mutate(Label=factor(Label,levels = rev(c("High","Low","N.S.")))), aes(x=Count,y=Drugs,fill=Label))+ geom_col()+ scale_fill_manual(values = rev(c("#E64B35FF","#3C5488FF","#B09C85FF")))+ labs(x="No. of cancer",y="")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) p<-p1 %>% aplot::insert_right(p2,width = 0.3) ggsave(p,filename = "Drugs.PRISM.Drugs.Predicted.AUC.pdf",height = 6,width = 9) ###### CTRP ####### setwd("K:/TCGA/Anlysis/BCAA/Depmap") files <- list.files(".",pattern = "csv$") files <- files[str_detect(files,"CTRP")] RiskScore.Cancer <- read.csv(file = "../RiskGroup.COX.TCGA.RiskScore.csv") %>% group_by(Cancer) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group=if_else(total_risk_score >= cutpoint,"High","Low")) %>% ungroup() CTRP.RiskScore <- data.frame() for (file in files) { project = (file %>% str_split("\\.") %>% unlist())[3] print(project) if (project %in% RiskScore.Cancer$Cancer) { data <- read.csv(file,row.names = 1,check.names = F) %>% magrittr::set_colnames(str_remove_all(colnames(.),"\\(.*")) %>% merge(RiskScore.Cancer,by.x="V1",by.y="SampleID") High.G <- data %>% filter(Group == "High") Low.G <- data %>% filter(Group == "Low") for (drug in colnames(data)[2:407]) { if (length(na.omit(data[[drug]])) > 0) { test <- wilcox.test(High.G[[drug]],Low.G[[drug]],alternative = "two.sided", exact = F) CTRP.RiskScore <- data.frame(Pvalue=test$p.value, High.Mean=mean(High.G[[drug]]), High.Median=median(High.G[[drug]]), Low.Mean=mean(Low.G[[drug]]), Low.Median=median(Low.G[[drug]]), Drugs = drug, Cancer=project) %>% rbind.data.frame(CTRP.RiskScore) } } } } write.csv(CTRP.RiskScore,"Drugs.Depmap.CTRP.Wilcox.csv",row.names = F) CTRP.RiskScore.S <- CTRP.RiskScore %>% filter(Pvalue <= 0.05) %>% mutate(Label = if_else(High.Median>Low.Median,"High","Low")) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% arrange(desc(Count)) %>% mutate(Drugs=str_remove_all(Drugs," ")) #### Figure #### CTRP.RiskScore.S <- CTRP.RiskScore %>% #filter(Pvalue <= 0.05) %>% mutate(Label = if_else(Pvalue>0.05,"N.S.", if_else(High.Median>Low.Median,"High","Low"))) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% arrange(desc(Count))%>% filter(Label!="N.S.") %>% mutate(Drugs=str_remove_all(Drugs," ")) %>% filter(Count > 12) CTRP.RiskScore2 <- CTRP.RiskScore %>% mutate(Drugs=str_remove_all(Drugs," ")) %>% mutate(Label = if_else(Pvalue>0.05,"N.S.", if_else(High.Median>Low.Median,"High","Low"))) %>% filter(Drugs %in% c(CTRP.RiskScore.S$Drugs,"axitinib", "docetaxel:tanespimycin","linsitinib","sorafenib")) CTRP.RiskScore3 <- CTRP.RiskScore %>% #filter(Pvalue <= 0.05) %>% mutate(Label = if_else(Pvalue>0.05,"N.S.", if_else(High.Median>Low.Median,"High","Low"))) %>% group_by(Drugs,Label) %>% summarise(Count=n()) %>% arrange(desc(Count))%>% #filter(Label!="N.S.") %>% mutate(Drugs=str_remove_all(Drugs," ")) %>% filter(Drugs %in% c(CTRP.RiskScore.S$Drugs,"axitinib", "docetaxel:tanespimycin","linsitinib","sorafenib")) p1 <- ggplot(data=CTRP.RiskScore2, aes(x=Cancer,y=Drugs,color=Label,shape=Label) )+ geom_point(size=2)+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_npg()+ scale_color_manual(values = c("#E64B35FF","#3C5488FF","#B09C85FF"))+ scale_shape_manual(values = c(1,3,4))+ labs(x="",y="",color="Predicted AUC\nfor risk group", shape="Predicted AUC\nfor risk group") p2 <- ggplot(data=CTRP.RiskScore3 %>% mutate(Label=factor(Label,levels = rev(c("High","Low","N.S.")))), aes(x=Count,y=Drugs,fill=Label))+ geom_col()+ scale_fill_manual(values = rev(c("#E64B35FF","#3C5488FF","#B09C85FF")))+ labs(x="No. of cancer",y="")+ ggthemes::theme_few()+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid")) p<-p1 %>% aplot::insert_right(p2,width = 0.3) ggsave(p,filename = "Drugs.CTRP.Drugs.Predicted.AUC.pdf",height = 6,width = 9) CTRP <- read.csv("Drugs.Depmap.CTRP.Wilcox.csv") %>% filter(Pvalue < 0.05) %>% mutate(Database="CTRP") PRISM <- read.csv("Drugs.Depmap.PRISM.Wilcox.csv") %>% filter(Pvalue < 0.05) %>% mutate(Database="PRISM") GDSC2 <- read.csv("../Pancancer.GDSC2.Drugs.Sensitivity.csv") %>% filter(p.adj < 0.05) %>% mutate(Database="GDSC2") write.csv(GDSC2,file="GDSC2.csv") rbind(CTRP,PRISM) %>% write.csv("TableS22.csv") #################### ###### Heterogeneity => T cell ####### setwd("K:/TCGA/ZhangZeMin.Pancancer.T") library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) # AUCell、UCell和singscore + ssGSEA BCAA.ID <- read.csv("../Anlysis/BCAA/BCAA.ID.csv") files <- list.files(".",pattern = ".gz$") files <- files[str_detect(files,"CD8")|str_detect(files,"CD4")] metadata <- fread("GSE156728_metadata.txt.gz") CD4.CD8.Bulk.BCAA <- data.frame() CD4.CD8.scRNA.ssGSEA.BCAA <- data.frame() CD4.CD8.scRNA.AddModuleScore.BCAA <- data.frame() library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) for (file in files) { print(file) project=(file %>%str_remove_all("\\..*") %>%str_split("_") %>% unlist())[2] Cell <- (file %>%str_split("\\.") %>% unlist())[2] data1 <- fread(file) %>% data.frame() %>% #filter(V1 %in% BCAA.ID$SYMBOL) %>% remove_rownames() %>% column_to_rownames("V1") %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("cellID") %>% merge(metadata %>% dplyr::select(cellID,loc,meta.cluster,cancerType), by="cellID") %>% column_to_rownames("cellID") %>% dplyr::select(-cancerType) data2 <- aggregate(.~loc+meta.cluster,mean,data=data1) %>% mutate(axis=paste(loc,meta.cluster,sep = "_")) %>% dplyr::select(-loc,-meta.cluster) %>% remove_rownames() %>% column_to_rownames("axis") %>% t() %>% data.frame(check.names = F) library(GSVA) BCAA=list() BCAA[["BCAA"]]<-BCAA.ID$SYMBOL #data2 <- fread(file) %>% remove_rownames() %>% column_to_rownames("V1") res.ssgsea <- gsva(as.matrix(data2), gset.idx.list=BCAA, method = "ssgsea", kcdf = "Poisson") CD4.CD8.Bulk.BCAA <- data.frame(cellType=colnames(res.ssgsea), BCAA=res.ssgsea[1,],Cancer=project) %>% rbind.data.frame(CD4.CD8.Bulk.BCAA) data2 <- fread(file) %>% remove_rownames() %>% column_to_rownames("V1") res.ssgsea <- gsva(as.matrix(data2), gset.idx.list=BCAA, method = "ssgsea", kcdf = "Poisson") CD4.CD8.scRNA.ssGSEA.BCAA <- data.frame(cellType=colnames(res.ssgsea), BCAA=res.ssgsea[1,],Cancer=project) %>% rbind.data.frame(CD4.CD8.scRNA.ssGSEA.BCAA) sce <- CreateSeuratObject(data2) sce[["percent.mt"]] <- PercentageFeatureSet( sce, pattern = "^MT-" ) meta <- metadata %>% filter(cancerType==project) #meta<-data.table::fread("GSE210347_meta.txt",data.table = F) %>% # dplyr::select(cellname,SampleID,seurat_clusters,cluster,celltype,tissue,group) data <- sce@meta.data data <- rownames_to_column(data,var = "barcodes") meta1 <- merge(data, meta, by.x='barcodes', by.y='cellID', all=T) meta1 <- column_to_rownames(meta1,var = "barcodes") sce <- AddMetaData(sce,meta1) sce %<>% subset(subset = nFeature_RNA > 200 & nFeature_RNA < 6000 & percent.mt < 20) %>% #nolint NormalizeData(normalization.method = "LogNormalize") %>% FindVariableFeatures(selection.method = "vst") sce %<>% ScaleData() sce %<>% RunPCA(object = ., features = VariableFeatures(object = .)) sce <- RunUMAP(sce, dims = 1:30, reduction = "pca") #sce <- AddMetaData(sce,metadata = cancer.metadata) p1 <- DimPlot(sce,group.by = "meta.cluster", split.by = "loc",cols = color)+ theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"), plot.title = element_blank()) ggsave(p1,filename = paste(project,Cell,"loc.cluster.pdf",sep = "."),width = 12,height = 4) p1 <- DimPlot(sce,group.by = "meta.cluster",cols = color)+ theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"), plot.title = element_blank()) ggsave(p1,filename = paste(project,Cell,"cluster.pdf",sep = "."),width = 8,height = 4) p1 <- FeaturePlot(sce,features = c("BCAT1"))+ theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = c("grey90",brewer.pal(11,"RdBu")[5:1])) ggsave(p1,filename = paste(project,Cell,"BCAT1.pdf",sep = "."),height = 3,width = 3.5) p1 <- FeaturePlot(sce,features = c("BCAT2"))+ theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = c("grey90",brewer.pal(11,"RdBu")[5:1])) ggsave(p1,filename = paste(project,Cell,"BCAT2.pdf",sep = "."),height = 3,width = 3.5) BCAA <- c("ACAD8","ACADSB","ACAT1","ALDH6A1","AUH","BCAT1","BCAT2","BCKDHA","BCKDHB","DBT","DLD","ECHS1","HIBADH","HIBCH","HSD17B10","IVD","MCCC1","MCCC2") gene <- as.list(BCAA) sce <- AddModuleScore(sce,features = gene,name = 'BCAA') saveRDS(sce, paste(project,Cell,"sce.Preprocess.Rds",sep = ".")) p1 <- FeaturePlot(sce,features = c("BCAA1"))+ theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = c("grey90",brewer.pal(11,"RdBu")[5:1]))+ labs(title="BCAA score") ggsave(p1,filename = paste(project,Cell,"BCAA.score.pdf",sep = "."),height = 3,width = 3.5) mydata<- FetchData(sce,vars = c("BCAA1")) %>% mutate(Cancer=project) #write.csv(mydata,"FetchData.UMAP.BCAA.csv") CD4.CD8.scRNA.AddModuleScore.BCAA <- rbind.data.frame(CD4.CD8.scRNA.AddModuleScore.BCAA,mydata) } #CD4.CD8.Bulk.BCAA <- data.frame() #CD4.CD8.scRNA.ssGSEA.BCAA <- data.frame() #CD4.CD8.scRNA.AddModuleScore.BCAA <- data.frame() write.csv(CD4.CD8.Bulk.BCAA,file = "Pancancer.CD4.CD8.Bulk.BCAA.csv",row.names = F) write.csv(CD4.CD8.scRNA.ssGSEA.BCAA,file = "Pancancer.CD4.CD8.scRNA.ssGSEA.BCAA",row.names = F) write.csv(CD4.CD8.scRNA.AddModuleScore.BCAA,file = "Pancancer.CD4.CD8.scRNA.AddModuleScore.BCAA.csv",row.names = F) #### Figure #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/T") data <- read.csv("Pancancer.CD4.CD8.Bulk.BCAA.csv") %>% mutate(CellType=str_remove(cellType,".*_")) %>% mutate(Loc = str_remove(cellType,"_.*")) %>% mutate(Yaxis=paste(CellType,Loc,sep = "_")) p1 <- ggplot(data,aes(y=Cancer,x=Yaxis))+ geom_tile(aes(fill=BCAA),color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradientn(colors = brewer.pal(9,"Reds")) # YlOrRd Blues BuGn library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) mid <- data %>% dplyr::select(Yaxis,CellType) %>% unique() %>% group_by(CellType) %>% summarise(Count=n()) p2 <- ggplot(data %>% dplyr::select(Yaxis,CellType) %>% unique(), aes(y=1,x=Yaxis))+ geom_col(aes(fill=CellType),width = 1)+ #ggthemes::theme_few()+ #theme_void()+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ scale_fill_manual(values = color)+ labs(x="",y="") #scale_fill_gradientn(colors = brewer.pal(9,"Reds")) # YlOrRd Blues BuGn mid <- data %>% dplyr::select(Loc,CellType) %>% unique() %>% group_by(CellType) %>% summarise(Count=n()) p3 <- ggplot(data %>% dplyr::select(Yaxis,Loc) %>% unique(), aes(y=1,x=Yaxis))+ geom_col(aes(fill=Loc),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") #scale_fill_manual(values = color) p123 <- p1 %>% aplot::insert_top(p2,height = 0.1) %>% aplot::insert_top(p3,height = 0.1) ggsave(p123,filename = "Hetero.T.BCAAscore.ssGSEA.pdf",height = 3,width = 18) ggsave(p123,filename = "Hetero.T.BCAAscore.ssGSEA.Legend.pdf",height = 10,width = 12) #### ESCA #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/T") data <- read.csv("Pancancer.CD4.CD8.Bulk.BCAA.csv") %>% mutate(CellType=str_remove(cellType,".*_")) %>% mutate(Loc = str_remove(cellType,"_.*")) %>% mutate(Yaxis=paste(CellType,Loc,sep = "_")) %>% filter(Cancer=="ESCA") p1 <- ggplot(data %>% mutate(Loc=factor(Loc,levels = c("T","N"))) ,aes(x=BCAA,y=Yaxis))+ geom_segment(aes(x=0,xend=BCAA,y=Yaxis,yend=Yaxis),color="black")+ geom_point(aes(color=Loc),size=3)+ ggthemes::theme_few()+ theme(legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=7,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=8), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="BCAA score",y="")+ geom_hline(yintercept = seq(2,100,2)+0.5,linetype=2)+ scale_color_npg() ggsave(p1,filename = "ESCA.BCAAscore.pdf",height = 8,width = 4) #### BCAA score => AUCell #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/T") files <- list.files(".",pattern = "AUCell.*.csv") AUCell.Stat <- data.frame() for (file in files) { data <- read.csv(file) %>% mutate(CellType=str_extract(meta.cluster,"\\..*\\.")) %>% mutate(CellType=str_remove_all(CellType,"\\.c.?.?\\.") %>% str_remove_all("\\.")) string <- file %>% str_split("\\.") %>% unlist() project <- string[2] Cell <- string[3] print(project) if (project != "FTC") { if (project %in% c("BCL","MM")) { middata <- data %>% group_by(loc,CellType) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(loc,Count) %>% filter(P>=3) %>% filter(T>=3) AUCell.Stat <- data %>% filter(CellType %in% middata$CellType) %>% group_by(CellType) %>% rstatix::t_test(AUC~loc,detailed = T) %>% mutate(Cancer=project,Cell=Cell) %>% plyr::rbind.fill(AUCell.Stat) }else{ middata <- data %>% group_by(loc,CellType) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(loc,Count) %>% filter(N>=3) %>% filter(T>=3) AUCell.Stat <- data %>% filter(CellType %in% middata$CellType) %>% group_by(CellType) %>% rstatix::t_test(AUC~loc,detailed = T) %>% mutate(Cancer=project,Cell=Cell) %>% plyr::rbind.fill(AUCell.Stat) } } } write.csv(AUCell.Stat,"AUCell.BCAA.T.Stat.csv",row.names = F) #### BCAA SYMBOLs => T test #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/T") data <- read.csv("BCAA.SYMBOLS.T.Expression.csv") %>% mutate(CellType=str_extract(meta.cluster,"\\..*\\.")) %>% mutate(CellType=str_remove_all(CellType,"\\.c.?.?\\.") %>% str_remove_all("\\.")) %>% mutate(CellType=paste(FCell,CellType,sep = "_")) SYMBOLS.Stat <- data.frame() for (project in unique(data$Cancer)) { #data <- read.csv(file) #string <- file %>% str_split("\\.") %>% unlist() #project <- string[2] #Cell <- string[3] print(project) mid <- data %>% filter(Cancer==project) %>% gather(SYMBOL,TPM,ACAD8:MCCC2) if (project != "FTC") { if (project %in% c("BCL","MM")) { middata <- mid %>% group_by(loc,CellType,SYMBOL) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(loc,Count) %>% filter(P>=3) %>% filter(T>=3) SYMBOLS.Stat <- mid %>% filter(CellType %in% middata$CellType) %>% group_by(CellType,SYMBOL) %>% rstatix::t_test(TPM~loc,detailed = T) %>% mutate(Cancer=project) %>% plyr::rbind.fill(SYMBOLS.Stat) }else{ middata <- mid %>% group_by(loc,CellType,SYMBOL) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(loc,Count) %>% filter(N>=3) %>% filter(T>=3) SYMBOLS.Stat <- mid %>% filter(CellType %in% middata$CellType) %>% group_by(CellType,SYMBOL) %>% rstatix::t_test(TPM~loc,detailed = T) %>% mutate(Cancer=project) %>% plyr::rbind.fill(SYMBOLS.Stat) } } } write.csv(SYMBOLS.Stat,"SYMBOLS.BCAA.T.Stat.csv",row.names = F) ###### Heterogeneity => T cell => CMP ###### setwd("K:/TCGA/ZhangZeMin.Pancancer.T") library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) # AUCell、UCell和singscore + ssGSEA countToFpkm <- function(counts, effLen) { N <- sum(counts) exp( log(counts) + log(1e9) - log(effLen) - log(N) ) } #Gene.Length <- read.csv("../gene.length.txt",sep = "\t",header = F) library(vegan) #BCAA.ID <- read.csv("../Anlysis/BCAA/BCAA.ID.csv") files <- list.files(".",pattern = ".gz$") files <- files[str_detect(files,"CD8")|str_detect(files,"CD4")] metadata <- fread("GSE156728_metadata.txt.gz") for (file in files) { print(file) project=(file %>%str_remove_all("\\..*") %>%str_split("_") %>% unlist())[2] Cell <- (file %>%str_split("\\.") %>% unlist())[2] if (!file.exists(paste(project,Cell,"CPM.csv",sep = "."))) { data1 <- fread(file) %>% data.frame() %>% #filter(V1 %in% BCAA.ID$SYMBOL) %>% remove_rownames() %>% column_to_rownames("V1") #%>% #t() %>% data.frame(check.names = F) %>% #rownames_to_column("cellID") %>% #merge(metadata %>% dplyr::select(cellID,loc,meta.cluster,cancerType), # by="cellID") %>% #column_to_rownames("cellID") %>% dplyr::select(-cancerType) #data <- vegan::decostand(data1,method = "total",MARGIN = 2) * 1e6 cpm <- apply(data1 ,2, function(x) { x/sum(x)*1000000 }) write.csv(data,file = paste(project,Cell,"CPM.csv",sep = ".")) } } #### Datasets #### setwd("K:/TCGA/ZhangZeMin.Pancancer.T") library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) BCAA.ID <- read.csv("../Anlysis/BCAA/BCAA.ID.csv") files <- list.files("CPM/",pattern = "csv$") files <- files[str_detect(files,"CPM") & !str_detect(files,"Pancancer")] metadata <- fread("GSE156728_metadata.txt.gz") BCAA.T <- data.frame() for (file in files) { print(file) #BCAA.T <- data.frame() BCAA.T <- read.csv(file.path("CPM",file),row.names = 1) %>% rownames_to_column("SYMBOL") %>% filter(SYMBOL %in% BCAA.ID$SYMBOL) %>% t() %>% data.frame(check.names = F) %>% rownames_to_column("cellID") %>% merge(metadata,by="cellID") %>% rbind.data.frame(BCAA.T) } write.csv(BCAA.T,"Pancancer.BCAA.SYMBOL.CPM.csv",row.names = F) #### Figure => Data => AveExp + PerExp #### setwd("K:/TCGA/Anlysis/BCAA/") BCAA.ID <- read.csv("BCAA.ID.csv") %>% arrange(SYMBOL) BCAA.T <- read.csv("Pancancer.BCAA.SYMBOL.CPM.csv") library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) colnames(BCAA.T)[2:19]=BCAA.ID$SYMBOL BCAA.T.AveExp <- BCAA.T %>% filter(loc != "L") %>% gather(SYMBOL,Expression,ACAD8:MCCC2) %>% mutate(cell=str_remove_all(meta.cluster,"\\..*")) %>% mutate(type=str_extract(meta.cluster,"\\..*\\.") %>% str_remove_all("\\.c..\\.") %>% str_remove_all("\\.")) %>% mutate(cluster=paste(cell,type,sep = "_")) %>% group_by(cancerType,cluster,loc,SYMBOL) %>% summarise(AveExp = mean(Expression)) %>% ungroup() BCAA.T.PerExp <- BCAA.T %>% filter(loc != "L") %>% gather(SYMBOL,Expression,ACAD8:MCCC2) %>% mutate(cell=str_remove_all(meta.cluster,"\\..*")) %>% mutate(type=str_extract(meta.cluster,"\\..*\\.") %>% str_remove_all("\\.c..\\.") %>% str_remove_all("\\.")) %>% mutate(cluster=paste(cell,type,sep = "_")) %>% group_by(cancerType,cluster,loc,SYMBOL) %>% mutate(TotalCount=sum(Expression>=0)) %>% mutate(ExpCount = sum(Expression >0)) %>% dplyr::select(cancerType,loc,cluster,TotalCount,ExpCount) %>% unique() %>% mutate(PerExp = ExpCount*100/TotalCount) write.csv(BCAA.T.AveExp,file = "BCAA.SYMBOLS.T.CLUSTER.AVEEXP.csv",row.names = F) write.csv(BCAA.T.PerExp,file = "BCAA.SYMBOLS.T.CLUSTER.PEREXP.csv",row.names = F) #### Figure => CD4 #### library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) BCAA.T.D <- merge(BCAA.T.AveExp,BCAA.T.PerExp, by=c("cancerType","loc","cluster","SYMBOL")) %>% data.frame(check.names = F) %>% filter(str_detect(cluster,"CD4")) %>% #filter(SYMBOL=="BCAT1") %>% mutate(mid = paste(cancerType,loc,sep = "_")) %>% mutate(Xaxis=paste(cluster,loc,cancerType,sep = "_")) %>% arrange(cluster,mid) %>% mutate(Xaxis=factor(Xaxis,levels = unique(.$Xaxis))) p1 <- ggplot(BCAA.T.D, aes(x=Xaxis,y=SYMBOL,fill=log(AveExp)))+ geom_tile(color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradient2(high = brewer.pal(11,"RdBu")[1], low=brewer.pal(11,"RdBu")[11]) p2 <- ggplot(BCAA.T.D,aes(x=Xaxis,y=SYMBOL,fill=PerExp))+ geom_tile(color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradientn(colours = brewer.pal(9,"Blues")) p3 <- ggplot(BCAA.T.D %>% dplyr::select(Xaxis,cancerType) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=cancerType),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ ggthemes::scale_fill_stata()+ #scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p4 <- ggplot(BCAA.T.D %>% dplyr::select(Xaxis,cluster) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=cluster),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ #ggthemes::scale_fill_stata()+ scale_fill_manual(values = color)+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p5 <- ggplot(BCAA.T.D %>% dplyr::select(Xaxis,loc) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=loc),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ #ggthemes::scale_fill_stata()+ scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p12345 <- p1 %>% aplot::insert_bottom(p2,height = 1) %>% aplot::insert_top(p3,height = 0.1)%>% aplot::insert_top(p4,height = 0.1)%>% aplot::insert_top(p5,height = 0.1) ggsave(p12345,filename = "Htero.SYMBOL.T.CD4.pdf",height = 8,width = 14) #### Figure => CD8 #### BCAA.T.D <- merge(BCAA.T.AveExp,BCAA.T.PerExp, by=c("cancerType","loc","cluster","SYMBOL")) %>% data.frame(check.names = F) %>% filter(str_detect(cluster,"CD8")) %>% #filter(SYMBOL=="BCAT1") %>% mutate(mid = paste(cancerType,loc,sep = "_")) %>% mutate(Xaxis=paste(cluster,loc,cancerType,sep = "_")) %>% arrange(cluster,mid) %>% mutate(Xaxis=factor(Xaxis,levels = unique(.$Xaxis))) p1 <- ggplot(BCAA.T.D, aes(x=Xaxis,y=SYMBOL,fill=log(AveExp)))+ geom_tile(color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradient2(high = brewer.pal(11,"RdBu")[1], low=brewer.pal(11,"RdBu")[11]) p2 <- ggplot(BCAA.T.D,aes(x=Xaxis,y=SYMBOL,fill=PerExp))+ geom_tile(color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradientn(colours = brewer.pal(9,"Blues")) p3 <- ggplot(BCAA.T.D %>% dplyr::select(Xaxis,cancerType) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=cancerType),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ ggthemes::scale_fill_stata()+ #scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p4 <- ggplot(BCAA.T.D %>% dplyr::select(Xaxis,cluster) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=cluster),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ #ggthemes::scale_fill_stata()+ scale_fill_manual(values = color)+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p5 <- ggplot(BCAA.T.D %>% dplyr::select(Xaxis,loc) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=loc),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ #ggthemes::scale_fill_stata()+ scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p12345 <- p1 %>% aplot::insert_bottom(p2,height = 1) %>% aplot::insert_top(p3,height = 0.1)%>% aplot::insert_top(p4,height = 0.1)%>% aplot::insert_top(p5,height = 0.1) ggsave(p12345,filename = "Htero.SYMBOL.T.CD8.pdf",height = 8,width = 14) #### Example => CD4 #### BCAA.T.D <- merge(BCAA.T.AveExp,BCAA.T.PerExp, by=c("cancerType","loc","cluster","SYMBOL")) %>% data.frame(check.names = F) %>% filter(str_detect(cluster,"CD4")) %>% #filter(SYMBOL=="BCAT1") %>% mutate(mid = paste(cancerType,loc,sep = "_")) %>% mutate(Xaxis=paste(cluster,loc,cancerType,sep = "_")) %>% arrange(cluster,mid) %>% mutate(Xaxis=factor(Xaxis,levels = unique(.$Xaxis))) Exa.BCAT1 <- BCAA.T.D %>% filter(SYMBOL == "BCAT1") %>% mutate(Xaxis=paste(cancerType,loc,sep = "_")) p1 <- ggplot(Exa.BCAT1,aes(x=Xaxis,y=cluster, color=log10(AveExp),size=PerExp))+ geom_point()+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ scale_color_gradient2(high = brewer.pal(11,"RdBu")[1], low=brewer.pal(11,"RdBu")[11],mid = "white")+ scale_size_continuous(breaks=c(25,50,75,100))+ labs(x="",y="") ggsave(p1,filename = "Htero.SYMBOL.T.CD4.BCAT1.pdf",height = 4,width = 6) #### BCAT1 => CD8 #### BCAA.T.D <- merge(BCAA.T.AveExp,BCAA.T.PerExp, by=c("cancerType","loc","cluster","SYMBOL")) %>% data.frame(check.names = F) %>% filter(str_detect(cluster,"CD8")) %>% #filter(SYMBOL=="BCAT1") %>% mutate(mid = paste(cancerType,loc,sep = "_")) %>% mutate(Xaxis=paste(cluster,loc,cancerType,sep = "_")) %>% arrange(cluster,mid) %>% mutate(Xaxis=factor(Xaxis,levels = unique(.$Xaxis))) Exa.BCAT1 <- BCAA.T.D %>% filter(SYMBOL == "BCAT1") %>% mutate(Xaxis=paste(cancerType,loc,sep = "_")) p1 <- ggplot(Exa.BCAT1,aes(x=Xaxis,y=cluster, color=log10(AveExp),size=PerExp))+ geom_point()+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ scale_color_gradient2(high = brewer.pal(11,"RdBu")[1], low=brewer.pal(11,"RdBu")[11],mid = "white")+ scale_size_continuous(breaks=c(25,50,75,100))+ labs(x="",y="") ggsave(p1,filename = "Htero.SYMBOL.T.CD8.BCAT1.pdf",height = 4,width = 6) ###### Heterogeneity => myeloid cell ####### setwd("K:/TCGA/ZhangZeMin.Pancancer.Myeloid/") library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) # AUCell、UCell和singscore + ssGSEA BCAA.ID <- read.csv("../Anlysis/BCAA/BCAA.ID.csv") files <- list.files(".",pattern = ".gz$") files <- files[str_detect(files,"expression")] library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) #metadata <- fread("GSE156728_metadata.txt.gz") Bulk.BCAA <- data.frame() for (file in files) { print(file) project=(file %>% str_split("_") %>% unlist())[2] metadata <- fread(paste("GSE154763_",project,"_metadata.csv.gz",sep = "")) #Cell <- (file %>%str_split("\\.") %>% unlist())[2] data1 <- fread(file) %>% data.frame() %>% #filter(V1 %in% BCAA.ID$SYMBOL) %>% #remove_rownames() %>% column_to_rownames("index") %>% #t() %>% data.frame(check.names = F) %>% #rownames_to_column("cellID") %>% merge(metadata %>% dplyr::select(index,tissue,MajorCluster), by="index") %>% column_to_rownames("index") data2 <- aggregate(.~tissue+MajorCluster,mean,data=data1) %>% mutate(axis=paste(tissue,MajorCluster,sep = "_")) %>% dplyr::select(-tissue,-MajorCluster) %>% remove_rownames() %>% column_to_rownames("axis") %>% t() %>% data.frame(check.names = F) library(GSVA) BCAA=list() BCAA[["BCAA"]]<-BCAA.ID$SYMBOL #data2 <- fread(file) %>% remove_rownames() %>% column_to_rownames("V1") res.ssgsea <- gsva(as.matrix(data2), gset.idx.list=BCAA, method = "ssgsea", kcdf = "Gaussian") Bulk.BCAA <- data.frame(cellType=colnames(res.ssgsea), BCAA=res.ssgsea[1,],Cancer=project) %>% rbind.data.frame(Bulk.BCAA) data2 <- fread(file) %>% remove_rownames() %>% column_to_rownames("index") %>% t() res.ssgsea <- gsva(as.matrix(data1), gset.idx.list=BCAA, method = "ssgsea", kcdf = "Gaussian") scRNA.ssGSEA.BCAA <- data.frame(cellType=colnames(res.ssgsea), BCAA=res.ssgsea[1,],Cancer=project) %>% rbind.data.frame(scRNA.ssGSEA.BCAA) sce <- CreateSeuratObject(data1) sce[["percent.mt"]] <- PercentageFeatureSet( sce, pattern = "^MT-" ) #meta <- metadata %>% filter(cancerType==project) metadata <- fread(paste("GSE154763_",project,"_metadata.csv.gz",sep = "")) #meta<-data.table::fread("GSE210347_meta.txt",data.table = F) %>% # dplyr::select(cellname,SampleID,seurat_clusters,cluster,celltype,tissue,group) data <- sce@meta.data data <- rownames_to_column(data,var = "barcodes") meta1 <- merge(data, meta, by.x='barcodes', by.y='index', all=T) meta1 <- column_to_rownames(meta1,var = "barcodes") sce <- AddMetaData(sce,meta1) sce %<>% #subset(subset = nFeature_RNA > 200 & nFeature_RNA < 6000 & percent.mt < 20) %>% #nolint NormalizeData(normalization.method = "LogNormalize") %>% FindVariableFeatures(selection.method = "vst") sce %<>% ScaleData() #sce %<>% RunPCA(object = ., features = VariableFeatures(object = .)) #sce <- RunUMAP(sce, dims = 1:30, reduction = "pca") #sce <- AddMetaData(sce,metadata = cancer.metadata) #p1 <- DimPlot(sce,group.by = "meta.cluster",split.by = "loc",cols = color)+ # theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"), # plot.title = element_blank()) #ggsave(p1,filename = paste(project,Cell,"loc.cluster.pdf",sep = "."),width = 12,height = 4) #p1 <- DimPlot(sce,group.by = "meta.cluster",cols = color)+ # theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"), # plot.title = element_blank()) #ggsave(p1,filename = paste(project,Cell,"cluster.pdf",sep = "."),width = 8,height = 4) #p1 <- FeaturePlot(sce,features = c("BCAT1"))+ # theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ # scale_color_gradientn(colors = c("grey90",brewer.pal(11,"RdBu")[5:1])) #ggsave(p1,filename = paste(project,Cell,"BCAT1.pdf",sep = "."),height = 3,width = 3.5) #p1 <- FeaturePlot(sce,features = c("BCAT2"))+ # theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ # scale_color_gradientn(colors = c("grey90",brewer.pal(11,"RdBu")[5:1])) #ggsave(p1,filename = paste(project,Cell,"BCAT2.pdf",sep = "."),height = 3,width = 3.5) #BCAA <- c("ACAD8","ACADSB","ACAT1","ALDH6A1","AUH","BCAT1","BCAT2","BCKDHA","BCKDHB","DBT","DLD","ECHS1","HIBADH","HIBCH","HSD17B10","IVD","MCCC1","MCCC2") #gene <- as.list(BCAA) #gene=list sce <- AddModuleScore(sce,features = BCAA,name = 'BCAA') saveRDS(sce, paste(project,Cell,"sce.Preprocess.Rds",sep = ".")) #p1 <- FeaturePlot(sce,features = c("BCAA1"))+ # theme(panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ # scale_color_gradientn(colors = c("grey90",brewer.pal(11,"RdBu")[5:1]))+ # labs(title="BCAA score") #ggsave(p1,filename = paste(project,Cell,"BCAA.score.pdf",sep = "."),height = 3,width = 3.5) mydata<- FetchData(sce,vars = c("BCAA1")) %>% mutate(Cancer=project) #write.csv(mydata,paste("FetchData.UMAP.BCAA.csv")) scRNA.AddModuleScore.BCAA <- rbind.data.frame(scRNA.AddModuleScore.BCAA,mydata) } write.csv(Bulk.BCAA,file = "Pancancer.Myeloid.Bulk.BCAA.csv",row.names = F) write.csv(scRNA.ssGSEA.BCAA,file = "Pancancer.Myeloid.scRNA.ssGSEA.BCAA",row.names = F) write.csv(scRNA.AddModuleScore.BCAA,file = "Pancancer.Myeloid.scRNA.AddModuleScore.BCAA.csv",row.names = F) #### Figure #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/Myeloid") library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) files <- list.files(".",pattern = "csv$") files <- files[!str_detect(files,"Hetero")] BCAT1.BCAT2 <- data.frame() for (file in files) { project = file %>% str_remove_all("_.*") print(project) BCAT1.BCAT2 <- read.csv(file) %>% dplyr::select(Sub_Cluster,Global_Cluster,Tissue,Global_UMAP_1,Global_UMAP_2,BCAT1,BCAT2,CancerType) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAT1.BCAT2) } write.csv(BCAT1.BCAT2,"Hetero.Myeloid.BCAT1.BCAT2.csv",row.names = F) for (project in unique(BCAT1.BCAT2$Cancer)) { P1 <- ggplot(BCAT1.BCAT2 %>% arrange(BCAT1) %>% filter(Cancer==project) %>% filter(Tissue != "L"), aes(Global_UMAP_1,Global_UMAP_2, color=BCAT1))+ geom_point(size=1)+ #facet_wrap(~Tissue,ncol=3)+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="")+ #scale_color_gradientn(colors = brewer.pal(9,"Reds")) scale_color_gradientn(colours = brewer.pal(9,"Reds")) P2 <- ggplot(BCAT1.BCAT2 %>% arrange(BCAT1) %>% filter(Cancer==project) %>% filter(Tissue != "L"), aes(Global_UMAP_1,Global_UMAP_2, color=Global_Cluster))+ geom_point(size=0.5)+ facet_wrap(~Tissue,ncol=3)+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="")+ #scale_color_gradientn(colors = brewer.pal(9,"Reds")) scale_color_manual(values = color) library(cowplot) ggsave(plot_grid(P1, P2, ncol = 1, align = "v"), filename = paste("Hetero.",project,"Myeloid.BCAT1.Cluster.pdf",sep = ""),height =7,width = 6) } #### Percent Expressed + Average expression #### files <- list.files(".",pattern = "csv$") files <- files[!str_detect(files,"Hetero")] BCAT1.BCAT2.All <- data.frame() for (file in files) { project = file %>% str_remove_all("_.*") print(project) BCAT1.BCAT2.All <- read.csv(file) %>% dplyr::select(Sub_Cluster,Global_Cluster,Tissue,Global_UMAP_1,Global_UMAP_2,ACAD8:MCCC2,CancerType) %>% gather(SYMBOL,Expression,ACAD8:MCCC2) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAT1.BCAT2.All) } #BCAT1.BCAT2 <- read.csv("Hetero.Myeloid.BCAT1.BCAT2.csv") write.csv(BCAT1.BCAT2.All,"Hetero.BCAT1.BCAT2.All.csv",row.names = F) BCAT1.BCAT2.AveExp <- BCAT1.BCAT2.All %>% filter(Tissue != "L") %>% group_by(Cancer,Tissue,Global_Cluster,SYMBOL) %>% summarise(AveExp = mean(Expression)) %>% ungroup() BCAT1.BCAT2.PerExp <- BCAT1.BCAT2.All %>% filter(Tissue != "L") %>% group_by(Cancer,Tissue,Global_Cluster,SYMBOL) %>% mutate(TotalCount=sum(Expression>=0)) %>% mutate(ExpCount = sum(Expression >0)) %>% dplyr::select(Cancer,Tissue,Global_Cluster,TotalCount,ExpCount) %>% unique() %>% mutate(PerExp = ExpCount*100/TotalCount) BCAT1.BCAT2 <- merge(BCAT1.BCAT2.AveExp,BCAT1.BCAT2.PerExp, by=c("Cancer","Tissue","Global_Cluster","SYMBOL")) %>% data.frame(check.names = F) %>% #filter(SYMBOL=="BCAT1") %>% mutate(mid = paste(Cancer,Tissue,sep = "_")) %>% mutate(Xaxis=paste(Global_Cluster,Tissue,Cancer,sep = "_")) %>% arrange(Global_Cluster,mid) %>% mutate(Xaxis=factor(Xaxis,levels = unique(.$Xaxis))) p1 <- ggplot(BCAT1.BCAT2,aes(x=Xaxis,y=SYMBOL,fill=AveExp))+ geom_tile(color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradientn(colours = brewer.pal(9,"Reds")) p2 <- ggplot(BCAT1.BCAT2,aes(x=Xaxis,y=SYMBOL,fill=PerExp))+ geom_tile(color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradientn(colours = brewer.pal(9,"Blues")) p3 <- ggplot(BCAT1.BCAT2 %>% dplyr::select(Xaxis,Cancer) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=Cancer),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ ggthemes::scale_fill_stata()+ #scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p4 <- ggplot(BCAT1.BCAT2 %>% dplyr::select(Xaxis,Global_Cluster) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=Global_Cluster),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ #ggthemes::scale_fill_stata()+ scale_fill_manual(values = color)+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p5 <- ggplot(BCAT1.BCAT2 %>% dplyr::select(Xaxis,Tissue) %>% unique(), aes(y=1,x=Xaxis))+ geom_col(aes(fill=Tissue),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ #ggthemes::scale_fill_stata()+ scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ #geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") p12345 <- p1 %>% aplot::insert_bottom(p2,height = 1) %>% aplot::insert_top(p3,height = 0.1)%>% aplot::insert_top(p4,height = 0.1)%>% aplot::insert_top(p5,height = 0.1) ggsave(p12345,filename = "Htero.SYMBOL.Myeloid.pdf",height = 8,width = 14) #### Example => BCAT1 => LYM => cDC1 #### Exa.BCAT1 <- BCAT1.BCAT2 %>% filter(SYMBOL == "BCAT1") %>% mutate(Xaxis=paste(Cancer,Tissue,sep = "_")) p1 <- ggplot(Exa.BCAT1,aes(x=Xaxis,y=Global_Cluster, color=AveExp,size=PerExp))+ geom_point()+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ labs(x="",y="") ggsave(p1,filename = "Htero.SYMBOL.Myeloid.BCAT1.pdf",height = 4,width = 6) #### Percent Expressed + Average expression => Samples #### library(data.table) library(tidyverse) library(Seurat) library(SeuratObject) library(magrittr) files <- list.files(".",pattern = "csv$") files <- files[!str_detect(files,"Hetero") & !str_detect(files,"Pancancer")] BCAT1.BCAT2.All <- data.frame() for (file in files) { project = file %>% str_remove_all("_.*") print(project) BCAT1.BCAT2.All <- read.csv(file) %>% dplyr::select(Sub_Cluster,Global_Cluster,Sample,Tissue,Global_UMAP_1,Global_UMAP_2,ACAD8:MCCC2,CancerType) %>% gather(SYMBOL,Expression,ACAD8:MCCC2) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAT1.BCAT2.All) } #BCAT1.BCAT2 <- read.csv("Hetero.Myeloid.BCAT1.BCAT2.csv") write.csv(BCAT1.BCAT2.All,"Hetero.BCAT1.BCAT2.All.Samples.csv",row.names = F) BCAT1.BCAT2.AveExp <- BCAT1.BCAT2.All %>% filter(Tissue != "L") %>% group_by(Cancer,Sample,Tissue,Global_Cluster,SYMBOL) %>% summarise(AveExp = mean(Expression)) %>% ungroup() BCAT1.BCAT2.PerExp <- BCAT1.BCAT2.All %>% filter(Tissue != "L") %>% group_by(Cancer,Tissue,Sample,Global_Cluster,SYMBOL) %>% mutate(TotalCount=sum(Expression>=0)) %>% mutate(ExpCount = sum(Expression >0)) %>% dplyr::select(Cancer,Tissue,Global_Cluster,TotalCount,ExpCount) %>% unique() %>% mutate(PerExp = ExpCount*100/TotalCount) BCAT1.BCAT2 <- merge(BCAT1.BCAT2.AveExp,BCAT1.BCAT2.PerExp, by=c("Cancer","Tissue","Sample","Global_Cluster","SYMBOL")) %>% data.frame(check.names = F) %>% #filter(SYMBOL=="BCAT1") %>% mutate(mid = paste(Cancer,Tissue,sep = "_")) %>% mutate(Xaxis=paste(Global_Cluster,Tissue,Cancer,sep = "_")) %>% arrange(Global_Cluster,mid) %>% mutate(Xaxis=factor(Xaxis,levels = unique(.$Xaxis))) write.csv(BCAT1.BCAT2,"Hetero.BCAT1.BCAT2.All.Samples.AveExp.PerExp.csv",row.names = F) #### Example => BCAT1 => LYM => cDC1 #### Exa.BCAT1 <- BCAT1.BCAT2 %>% filter(SYMBOL == "BCAT1") %>% mutate(Xaxis=paste(Cancer,Tissue,sep = "_")) %>% filter(!Global_Cluster %in% c("Monolike","Myeloid")) p1 <- ggplot(Exa.BCAT1,aes(x=Cancer,y=AveExp))+ geom_boxplot(aes(color=Tissue))+ facet_wrap(~Global_Cluster,scales = "free_y",nrow = 2)+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ ggsci::scale_fill_npg()+ labs(x="") ggsave(p1,filename = "Htero.SYMBOL.Myeloid.BCAT1.Boxplot.pdf",height = 5,width = 7) #### BCAA score psudobulk => Figure #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/Myeloid/") data <- read.csv("Pancancer.Myeloid.Bulk.BCAA.csv") %>% mutate(CellType=str_remove(cellType,"N_")) %>% mutate(CellType=str_remove(CellType,"P_")) %>% mutate(CellType=str_remove(CellType,"T_")) %>% mutate(Loc = str_remove(cellType,"_.*")) %>% mutate(Yaxis=paste(CellType,Loc,sep = "_")) p1 <- ggplot(data,aes(y=Cancer,x=Yaxis))+ geom_tile(aes(fill=BCAA),color="white")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_blank(), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="")+ scale_fill_gradientn(colors = brewer.pal(9,"Reds")) # YlOrRd Blues BuGn library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) mid <- data %>% dplyr::select(Yaxis,CellType) %>% unique() %>% group_by(CellType) %>% summarise(Count=n()) p2 <- ggplot(data %>% dplyr::select(Yaxis,CellType) %>% unique(), aes(y=1,x=Yaxis))+ geom_col(aes(fill=CellType),width = 1)+ #ggthemes::theme_few()+ #theme_void()+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ scale_fill_manual(values = color)+ labs(x="",y="") #scale_fill_gradientn(colors = brewer.pal(9,"Reds")) # YlOrRd Blues BuGn mid <- data %>% dplyr::select(Loc,CellType) %>% unique() %>% group_by(CellType) %>% summarise(Count=n()) p3 <- ggplot(data %>% dplyr::select(Yaxis,Loc) %>% unique(), aes(y=1,x=Yaxis))+ geom_col(aes(fill=Loc),width = 1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9,angle = 90,vjust = 0.5,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text = element_blank(), axis.ticks = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #ggsci::scale_color_nejm()+ scale_fill_manual(values = rev(c("#DC0000FF","#F39B7FFF","#00A087FF")))+ theme(legend.position = "bottom")+ geom_vline(xintercept = accumulate(mid$Count,sum)+0.5)+ labs(x="",y="") #scale_fill_manual(values = color) p123 <- p1 %>% aplot::insert_top(p2,height = 0.1) %>% aplot::insert_top(p3,height = 0.1) ggsave(p123,filename = "Hetero.Myeloid.BCAAscore.ssGSEA.pdf",height = 3,width = 18) ggsave(p123,filename = "Hetero.Myeloid.BCAAscore.ssGSEA.Legend.pdf",height = 10,width = 12) #### BCAA score AUCell => Stat+Figure #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/Myeloid") files <- list.files(".",pattern = "^AUC.*.csv") files <- files[str_detect(files,"BCAAscore")] for (file in files) { project <- (file %>% str_split("\\.") %>% unlist())[2] print(project) data <- read.csv(file) %>% mutate(CellType=str_extract_all(MajorCluster,"_.*_")) %>% mutate(CellType=str_remove_all(CellType,"_")) p1 <- ggplot(data,aes(UMAP1,UMAP2,color=CellType))+ geom_point(size=1)+ facet_wrap(~tissue)+ ggthemes::theme_few()+ theme_test(12)+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="") ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=4))) library(RColorBrewer) p2 <- ggplot(data %>% arrange(AUC),aes(UMAP1,UMAP2,color=AUC))+ geom_point(size=1)+ facet_wrap(~tissue)+ ggthemes::theme_few()+ theme_test(12)+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="") scale_color_gradientn(colours = brewer.pal(9,"Reds")) library(cowplot) p <- plot_grid(p1, p2, ncol = 1, align = "v") ggsave(p,filename = paste("../../Hetero.BCAA.AUCell.Score.Myeloid",project,"Figure.pdf",sep = "."), height = 4,width = 6) } data <- read.csv("AUCell.KIDNEY.BCAAscore.csv") %>% mutate(CellType=str_extract_all(MajorCluster,"_.*_")) %>% mutate(CellType=str_remove_all(CellType,"_")) library(ggpubr) p123 <- ggplot(data%>% filter(CellType %in% c("cDC3","pDC")), aes(tissue,AUC,color=tissue))+ facet_wrap(~CellType)+ geom_boxplot(linetype="dashed",color="black")+ stat_boxplot(aes(ymin=..lower..,ymax=..upper.., fill=tissue), color="black")+ stat_boxplot(geom = "errorbar",aes(ymin=..ymax..,color=tissue),width=0.3,size=1)+ stat_boxplot(geom = "errorbar",aes(ymax=..ymin..,color=tissue),width=0.3,size=1)+ stat_compare_means(label = "p.format",label.x = 1.2,method = "t.test")+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(legend.position = "none", axis.text = element_text(color = "black"), #axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 #axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ theme(strip.background=element_rect(fill=c("#6F99AD7F", "#EE4C977F")))+ scale_fill_manual(values = c("#E64B35FF","#4DBBD5FF"))+ labs(x="") ggsave(p123,filename = "AUCell.KIDNEY.Example.Boxplot.pdf", height = 3.5,width = 2.5) files <- list.files(".",pattern = "^AUC.*.csv") AUCell.Stat <- data.frame() for (file in files) { project <- (file %>% str_split("\\.") %>% unlist())[2] print(project) data <- read.csv(file) %>% mutate(CellType=str_extract_all(MajorCluster,"_.*_")) %>% mutate(CellType=str_remove_all(CellType,"_")) if (project %in% c("LYM")) { middata <- data %>% group_by(tissue,CellType) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(tissue,Count) %>% filter(P>=3) %>% filter(T>=3) AUCell.Stat <- data %>% filter(CellType %in% middata$CellType) %>% group_by(CellType) %>% rstatix::t_test(AUC~tissue,detailed = T) %>% rbind.data.frame(AUCell.Stat) }else if (project %in% c("PAAD","THCA")){ middata <- data %>% group_by(tissue,CellType) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(tissue,Count) %>% filter(N>=3) %>% filter(T>=3) AUCell.Stat <- data %>% filter(CellType %in% middata$CellType) %>% group_by(CellType) %>% rstatix::t_test(AUC~tissue,detailed = T) %>% rbind.data.frame(AUCell.Stat) }else { AUCell.Stat <- data %>% #filter(CellType %in% middata$CellType) %>% group_by(CellType) %>% rstatix::t_test(AUC~tissue,detailed = T) %>% rbind.data.frame(AUCell.Stat) } } write.csv(AUCell.Stat,"AUCell.BCAA.Meyloid.Stat.csv",row.names = F) #### BCAA SYMBOLs => T test #### data <- read.csv("BCAA.SYMBOLS.Myeloid.Expression.csv") %>% mutate(CellType=str_extract_all(MajorCluster,"_.*_")) %>% mutate(CellType=str_remove_all(CellType,"_")) BCAA_SYMBOLS.Stat <- data.frame() for (project in unique(data$Cancer)) { mid <- data %>% filter(Cancer==project) %>% gather(SYMBOL,TPM,ACAD8:MCCC2) if (project %in% c("LYM")) { middata <- mid %>% group_by(tissue,CellType,SYMBOL) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(tissue,Count) %>% filter(P>=3) %>% filter(T>=3) BCAA_SYMBOLS.Stat <- mid %>% filter(CellType %in% middata$CellType) %>% group_by(SYMBOL,CellType) %>% rstatix::t_test(TPM~tissue,detailed = T) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAA_SYMBOLS.Stat) }else if (project %in% c("PAAD","THCA")){ middata <- mid %>% group_by(tissue,CellType,SYMBOL) %>% summarise(Count=n()) %>% arrange(CellType) %>% ungroup() %>% spread(tissue,Count) %>% filter(N>=3) %>% filter(T>=3) BCAA_SYMBOLS.Stat <- mid %>% filter(CellType %in% middata$CellType) %>% group_by(SYMBOL,CellType) %>% rstatix::t_test(TPM~tissue,detailed = T) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAA_SYMBOLS.Stat) }else { BCAA_SYMBOLS.Stat <- mid %>% #filter(CellType %in% middata$CellType) %>% group_by(SYMBOL,CellType) %>% rstatix::t_test(TPM~tissue,detailed = T) %>% mutate(Cancer=project) %>% rbind.data.frame(BCAA_SYMBOLS.Stat) } } write.csv(BCAA_SYMBOLS.Stat,file = "BCAA_SYMBOLS.Meyloid.Stat.csv",row.names = F) ###### Heterogeneity => CCLE ############## pkgs <- c( "Seurat", "SeuratWrappers", "ggplot2", "batchelor", "dplyr", "optparse", "reshape2", "data.table", "magrittr" ) lapply(pkgs, function(x) require(package = x, character.only = TRUE, quietly = TRUE, warn.conflicts = FALSE)) #nolint library(Seurat) library(ggplot2) library(clustree) library(cowplot) library(dplyr) suppressMessages(suppressWarnings(library(ggsci))) suppressMessages(suppressWarnings(library(harmony))) suppressMessages(suppressWarnings(library(scater))) suppressMessages(suppressWarnings(library(Seurat))) suppressMessages(suppressWarnings(library(tidyverse))) suppressMessages(suppressWarnings(library(readxl))) suppressMessages(suppressWarnings(library(SingleCellExperiment))) suppressMessages(suppressWarnings(library(scRNAseq))) suppressMessages(suppressWarnings(library(scran))) suppressMessages(suppressWarnings(library(Matrix))) suppressMessages(suppressWarnings(library(scater))) suppressMessages(suppressWarnings(library(ggthemes))) suppressMessages(suppressWarnings(library(cowplot))) data <- read.table("CPM_data.txt",header = T,row.names = 1) sce <- CreateSeuratObject(data) sce[["percent.mt"]] <- PercentageFeatureSet( sce, pattern = "^MT-" ) sce %<>% subset(subset = nFeature_RNA > 200 & nFeature_RNA < 6000 & percent.mt < 20) %>% #nolint NormalizeData(normalization.method = "LogNormalize") %>% FindVariableFeatures(selection.method = "vst") sce %<>% ScaleData() BCAA <- c("ACAD8","ACADSB","ACAT1","ALDH6A1","AUH","BCAT1","BCAT2","BCKDHA","BCKDHB","DBT","DLD","ECHS1","HIBADH","HIBCH","HSD17B10","IVD","MCCC1","MCCC2") gene <- as.list(BCAA) sce <- AddModuleScore(sce,features = gene,name = 'BCAA') mydata<- FetchData(sce,vars = c("BCAA1")) write.csv(mydata,"FetchData.BCAAscore.csv") BCAA.sce <- sce[BCAA,] saveRDS(BCAA.sce,file = "CCLE.BCAA.Rds") #### Figure #### setwd("K:/TCGA/Pan-cancer cell line heterogeneity") files <- list.files(".",pattern = "txt$") files <- files[str_detect(files,"tSNE")] Exp.Data <- read.csv("BCAA.CPM.csv",row.names = 1,check.names = F) %>% t() Metadata <- read.table("Metadata.txt",sep="\t",header = T,check.names = F) CCLE.SYMBOL=data.frame() for (file in files) { project=file %>% str_remove_all("tSNE_") %>% str_remove_all("\\..*") CCLE.SYMBOL <- read.table(file,skip = 1,sep = "\t",header = T) %>% set_colnames(c("NAME","tSNE1","tSNE2")) %>% na.omit() %>% merge(Metadata,by="NAME") %>% dplyr::select(NAME,tSNE1,tSNE2,Cell_line) %>% mutate(Cancer=project) %>% mutate(Cell_line = str_remove_all(Cell_line,"_.*")) %>% merge(Exp.Data %>% data.frame(check.names = F) %>% rownames_to_column("NAME") %>% mutate(NAME=str_replace_all(NAME,"\\.","-")),by="NAME") %>% rbind.data.frame(CCLE.SYMBOL) } write.csv(CCLE.SYMBOL,"CCLE.SYMBOLS.BCAA.CPM.csv",row.names = F) DesDir <- "../Anlysis/BCAA/Hetero/CCLE/" for (project in unique(CCLE.SYMBOL$Cancer)) { print(project) middata <- CCLE.SYMBOL %>% #filter(SYMBOL=="BCAT1") %>% filter(Cancer==project) p1 <- ggplot(middata %>% arrange(BCAT1),aes(tSNE1,tSNE2,color=BCAT1))+ geom_point(size=1)+ #facet_wrap(~Tissue,ncol=3)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="")+ #scale_color_gradientn(colors = brewer.pal(9,"Reds")) scale_color_gradientn(colours = brewer.pal(9,"Reds")[1:9]) p2 <- ggplot(middata %>% arrange(BCAT1), aes(tSNE1,tSNE2,color=Cell_line))+ geom_point(size=0.5)+ #facet_wrap(~Tissue,ncol=3)+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.y = element_text(size=9), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.x = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="")+ #scale_color_gradientn(colors = brewer.pal(9,"Reds")) scale_color_manual(values = color) library(cowplot) ggsave(p1 %>% aplot::insert_right(p2,width = 1), #plot_grid(p1, p2, ncol = 2, align = "h"), filename = paste(DesDir,"Hetero.",project,"CellLine.BCAT1.pdf",sep = ""), height =3,width = 7) } #### Cell_Line => SYMBOLs #### CCLE.SYMBOL <- read.csv("CCLE.SYMBOLS.BCAA.CPM.csv") CCLE.SYMBOL2 <- CCLE.SYMBOL %>% gather(SYMBOL,Exp,ACAD8:MCCC2) %>% filter(!Cancer %in% c("Fibroblast","Gallbladder_Cancer")) %>% group_by(Cancer,Cell_line,SYMBOL) %>% summarise(Exp.M = mean(Exp)) %>% ungroup() CCLE.SYMBOL2.Test <- CCLE.SYMBOL %>% gather(SYMBOL,Exp,ACAD8:MCCC2) %>% filter(!Cancer %in% c("Fibroblast","Gallbladder_Cancer")) %>% group_by(Cancer,SYMBOL) %>% rstatix::kruskal_test(Exp ~ Cell_line) %>% add_significance("p") CCLE.SYMBOL2.Final <- merge(CCLE.SYMBOL2,CCLE.SYMBOL2.Test,by=c("Cancer","SYMBOL")) %>% arrange(Cancer,Cell_line) %>% mutate(Cell_line = factor(Cell_line,levels = unique(.$Cell_line))) p1 <- ggplot()+ geom_tile(data = CCLE.SYMBOL2.Final %>% filter(!Cancer %in% c("Fibroblast","Gallbladder_Cancer")), aes(Cell_line,SYMBOL,fill=Exp.M), color="white")+ geom_point(data=CCLE.SYMBOL2.Final %>% filter(p.signif!="ns"), aes(Cell_line,SYMBOL,fill=Exp.M), shape=1,color="grey90",size=0.1)+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_blank(), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="")+ #scale_color_gradientn(colors = brewer.pal(9,"Reds")) scale_fill_gradientn(colours = brewer.pal(9,"Reds")[1:9])+ labs(x="",y="",fill="Ave Exp") library(rstatix) p2 <- ggplot(data = CCLE.SYMBOL2.Final %>% filter(!Cancer %in% c("Fibroblast","Gallbladder_Cancer")) %>% dplyr::select(Cancer,Cell_line) %>% unique(), aes(Cell_line,1,fill=Cancer))+ #geom_tile(color="white")+ geom_col(width = 1)+ #geom_text(aes(label=p.signif),color="white",angle=90)+ ggthemes::theme_few()+ theme(legend.position = "bottom", #axis.text = element_text(color = "black"), axis.text.x = element_blank(), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_blank(), axis.ticks.x = element_blank(), axis.ticks.y = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #labs(x="",y="")+ #scale_color_gradientn(colors = brewer.pal(9,"Reds")) #scale_fill_gradientn(colours = brewer.pal(9,"Reds")[1:9])+ scale_fill_manual(values = color)+ labs(x="",y="",color="Cancer") p12.2 <- p1 %>% aplot::insert_top(p2,height = 0.1) ggsave(p12.2,filename = paste(DesDir,"CCLE.SYMBOL.BCAA.pdf",sep = ""), height = 3,width = 18) #### BCAA score => Figure #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/CCLE") library(tidyverse) library(data.table) BCAA.ID <- read.csv("../../BCAA.ID.csv") metadata <- fread("Metadata.txt") data <- read.csv("CCLE.AUCell.SYMBOLS.csv") %>% dplyr::select(c("X","AUC")) %>% merge(metadata %>% mutate(NAME=str_replace_all(NAME,"-",".")), by.x="X",by.y="NAME") %>% mutate(celline=str_remove_all(Cell_line,"_.*")) AUCell.Stat <- data.frame() for (project in unique(data$Cancer_type)) { mid <- data %>% filter(Cancer_type == project) #%>% #gather(SYMBOL,TPM,ACAD8:MCCC2) if (length(unique(mid$celline)) > 1) { AUCell.Stat <- mid %>% #group_by(SYMBOL) %>% rstatix::t_test(AUC~celline,detailed = T) %>% mutate(Cancer=project) %>% plyr::rbind.fill(AUCell.Stat) } } write.csv(AUCell.Stat,file="AUCell.BCAA.CCLE.Stat.csv",row.names = F) metadata <- fread("Metadata.txt") data <- read.csv("CCLE.AUCell.SYMBOLS.csv") %>% dplyr::select(c("X","AUC")) %>% merge(metadata %>% mutate(NAME=str_replace_all(NAME,"-",".")), by.x="X",by.y="NAME") %>% mutate(celline=str_remove_all(Cell_line,"_.*")) # unique(data$Cancer_type) for (project in unique(data$Cancer_type)) { tsne.filename <- project %>% str_replace_all(" ","_") %>% str_replace_all("/","_") %>% paste("tSNE_",.,".txt",sep = "") figure.filename <- project %>% str_replace_all(" ","_") %>% str_replace_all("/","_") %>% paste("../../Hetero.BCAA.AUCell.CCLE.",.,".pdf",sep = "") tsne <- fread(tsne.filename,skip = 2) %>% mutate(V1=str_replace_all(V1,"-",".")) %>% merge(data,by.x="V1",by.y="X") print(project) print(dim(tsne)) print(length(unique(tsne$celline))) library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) if (length(unique(tsne$celline)) <= 70) { p1 <- ggplot(tsne %>% arrange(AUC), aes(V2,V3,color=AUC))+ geom_point(size=1)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ labs(x="tsne1",y="tsne2") p2 <- ggplot(tsne %>% arrange(AUC), aes(V2,V3,color=celline))+ geom_point(size=1)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ #scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ #ggthemes::scale_color_stata()+ scale_color_manual("Cell line",values = color)+ labs(x="Tsne1",y="Tsne2",color="Cell line")+ guides(color = guide_legend(override.aes = list(size=4))) p <- p1+p2 ggsave(p,filename = figure.filename,height = 3.5,width = 9) } } #### BCAA SYMBOLs => T test #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/CCLE") library(tidyverse) library(data.table) BCAA.ID <- read.csv("../../BCAA.ID.csv") metadata <- fread("Metadata.txt") data <- read.csv("CCLE.AUCell.SYMBOLS.csv") %>% dplyr::select(c("X",BCAA.ID$SYMBOL)) %>% merge(metadata %>% mutate(NAME=str_replace_all(NAME,"-",".")), by.x="X",by.y="NAME") %>% mutate(celline=str_remove_all(Cell_line,"_.*")) SYMBOLS.Stat <- data.frame() for (project in unique(data$Cancer_type)) { mid <- data %>% filter(Cancer_type == project) %>% gather(SYMBOL,TPM,ACAD8:MCCC2) if (length(unique(mid$celline)) > 1) { SYMBOLS.Stat <- mid %>% group_by(SYMBOL) %>% rstatix::t_test(TPM~celline,detailed = T) %>% mutate(Cancer=project) %>% plyr::rbind.fill(SYMBOLS.Stat) } } write.csv(SYMBOLS.Stat,file="BCAA_SYMBOLS.CCLE.Stat.csv",row.names = F) ######## Pancancer => scRNA ######### sce <- readRDS("sce.FindCluster.Rds") library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) sce2 <- subset(sce,celltype != "Plasma" & celltype != "undefined" & celltype != "Erythrocyte") #sce2$Axis <- paste(sce2$celltype,sce2$tissue,sce2$group,sep="_") sce2$Axis <- paste(sce2$celltype,sce2$tissue,sce2$group,sep="_") BCAA <- c("ACAD8","ACADSB","ACAT1","ALDH6A1","AUH","BCAT1","BCAT2","BCKDHA","BCKDHB","DBT","DLD","ECHS1","HIBADH","HIBCH","HSD17B10","IVD","MCCC1","MCCC2") p1 <- DotPlot(sce2, features = BCAA,assay='RNA',group.by ="Axis")+ coord_flip()+scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ scale_size_continuous(range = c(0.1,3))+ theme(axis.text.x = element_text(size=6,angle=90,vjust = 0.3,hjust = 1,color = "black"), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="") ggsave(p1,filename = "Dotplot.BCAA.SYMBOLS.pdf",height = 6,width = 14) sce3 <- subset(sce2,tissue == "PDAC") P1<-FeaturePlot(object = sce3, features = c("BCAT1"),reduction='umap', split.by = "group", cols = c('grey','red')) ggsave('Featureplot.PDAC.BCAT1.pdf',P1,height=4,width=11) sce4 <- subset(sce2,celltype == "Epithelium") sce4$Axis <- paste(sce4$cluster,sce4$tissue,sce4$group,sep="_") p1 <- DotPlot(sce4, features = BCAA,assay='RNA',group.by ="Axis")+ coord_flip()+scale_color_gradientn(colors = brewer.pal(9,"Reds"))+ scale_size_continuous(range = c(0.1,3))+ theme(axis.text.x = element_text(size=6,angle=90,vjust = 0.3,hjust = 1,color = "black"), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="") ggsave(p1,filename = "Dotplot.BCAA.SYMBOLS.Epithelium.pdf",height = 6,width = 14) #### BCAA SYMBOLs => T test #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/GSE210347") BCAA.ID <- read.csv("../../BCAA.ID.csv") files <- list.files(".",pattern = "AUCell") UMAP.D <- read.csv("embed_umap.csv") data.T <- data.frame() data.T2 <- data.frame() for (file in files) { mid <- read.csv(file) %>% dplyr::select(c("X",BCAA.ID$SYMBOL,cluster,celltype,group,tissue)) %>% gather(SYMBOL,TPM,ACAD8:MCCC2) %>% mutate(TPM=log2(TPM+1)) %>% filter(!celltype %in% c("Erythrocyte","undefined")) if (length(mid$group %>% unique()) > 1) { data.T <- mid %>% group_by(celltype,SYMBOL) %>% rstatix::t_test(TPM~group,detailed = T) %>% mutate(tissue=unique(mid$tissue)) %>% plyr::rbind.fill(data.T) } data.T2 <- mid %>% group_by(group,SYMBOL) %>% rstatix::t_test(TPM~celltype,detailed = T) %>% mutate(tissue=unique(mid$tissue)) %>% plyr::rbind.fill(data.T2) } write.csv(data.T,file = "GSE210347.SYMBOLS.Ttest.CellType.Loction.csv",row.names = F) write.csv(data.T2,file = "GSE210347.SYMBOLS.Ttest.Loction.CellType.csv",row.names = F) data.T2 <- data.frame() for (file in files) { mid <- read.csv(file) %>% dplyr::select(c("X",BCAA.ID$SYMBOL,cluster,celltype,group,tissue)) %>% gather(SYMBOL,TPM,ACAD8:MCCC2) %>% mutate(TPM=log2(TPM+1)) %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% filter(celltype == "Epithelium") mid.count <- mid %>% group_by(group,SYMBOL,cluster) %>% summarise(Count=n()) %>% data.frame() %>% dplyr::select(-SYMBOL) %>% unique() %>% filter(Count >= 3) data.T2 <- mid %>% merge(mid.count,by=c("cluster","group")) %>% group_by(group,SYMBOL) %>% rstatix::t_test(TPM~cluster,detailed = T) %>% mutate(tissue=unique(mid$tissue)) %>% plyr::rbind.fill(data.T2) } write.csv(data.T2,file = "GSE210347.SYMBOLS.Ttest.Epithelium.Loction.Cluster.csv",row.names = F) data <- read.csv("GSE210347.PDAC.AUCell.SYMBOLS.csv") %>% merge(UMAP.D,by="X") library(RColorBrewer) p1 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% arrange(BCAT1), aes(UMAP_1,UMAP_2,color=BCAT1))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds")) p2 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")),# aes(UMAP_1,UMAP_2,color=celltype))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=5))) library(patchwork) p <- p1/p2 ggsave(p,filename = "../../Hetero.GSE210347.SYMBOLS.BCAT1.PDAC.pdf", height = 7.5,width = 10) data <- read.csv("GSE210347.ICC.AUCell.SYMBOLS.csv")%>% merge(UMAP.D,by="X") library(RColorBrewer) p1 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% arrange(BCAT1), aes(UMAP_1,UMAP_2,color=BCAT1))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds")) p2 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")),# aes(UMAP_1,UMAP_2,color=celltype))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=5))) library(patchwork) p <- p1/p2 ggsave(p,filename = "../../Hetero.GSE210347.SYMBOLS.BCAT1.ICC.pdf", height = 7.5,width = 8.5) data <- read.csv("GSE210347.colorectal.AUCell.SYMBOLS.csv")%>% merge(UMAP.D,by="X") library(RColorBrewer) p1 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% arrange(BCAT1), aes(UMAP_1,UMAP_2,color=BCAT1))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds")) p2 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")),# aes(UMAP_1,UMAP_2,color=celltype))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=5))) library(patchwork) p <- p1/p2 ggsave(p,filename = "../../Hetero.GSE210347.SYMBOLS.BCAT1.Colorectal.pdf", height = 7.5,width = 10) #### BCAA score => AUCell => T test => Figure #### setwd("K:/TCGA/Anlysis/BCAA/Hetero/GSE210347") BCAA.ID <- read.csv("../../BCAA.ID.csv") files <- list.files(".",pattern = "AUCell") UMAP.D <- read.csv("embed_umap.csv") data.T <- data.frame() data.T2 <- data.frame() for (file in files) { mid <- read.csv(file) %>% dplyr::select(c("X",AUC,cluster,celltype,group,tissue)) %>% #gather(SYMBOL,TPM,ACAD8:MCCC2) %>% #mutate(TPM=log2(TPM+1)) %>% filter(!celltype %in% c("Erythrocyte","undefined")) if (length(mid$group %>% unique()) > 1) { data.T <- mid %>% group_by(celltype) %>% rstatix::t_test(AUC~group,detailed = T) %>% mutate(tissue=unique(mid$tissue)) %>% plyr::rbind.fill(data.T) } data.T2 <- mid %>% group_by(group) %>% rstatix::t_test(AUC~celltype,detailed = T) %>% mutate(tissue=unique(mid$tissue)) %>% plyr::rbind.fill(data.T2) } write.csv(data.T,file = "GSE210347.BCAA.AUCell.Ttest.CellType.Loction.csv",row.names = F) write.csv(data.T2,file = "GSE210347.BCAA.AUCell.Ttest.Loction.CellType.csv",row.names = F) data <- read.csv("GSE210347.PDAC.AUCell.SYMBOLS.csv") %>% merge(UMAP.D,by="X") library(RColorBrewer) p1 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% arrange(AUC), aes(UMAP_1,UMAP_2,color=AUC))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds")) p2 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")),# aes(UMAP_1,UMAP_2,color=celltype))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=5))) library(patchwork) p <- p1/p2 ggsave(p,filename = "../../Hetero.GSE210347.BCAA.AUCell.Score.PDAC.pdf", height = 7.5,width = 10) data <- read.csv("GSE210347.ICC.AUCell.SYMBOLS.csv")%>% merge(UMAP.D,by="X") library(RColorBrewer) p1 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% arrange(AUC), aes(UMAP_1,UMAP_2,color=AUC))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds")) p2 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")),# aes(UMAP_1,UMAP_2,color=celltype))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=5))) library(patchwork) p <- p1/p2 ggsave(p,filename = "../../Hetero.GSE210347.BCAA.AUCell.Score.ICC.pdf", height = 7.5,width = 8.5) data <- read.csv("GSE210347.colorectal.AUCell.SYMBOLS.csv")%>% merge(UMAP.D,by="X") library(RColorBrewer) p1 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")) %>% arrange(AUC), aes(UMAP_1,UMAP_2,color=AUC))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ scale_color_gradientn(colors = brewer.pal(9,"Reds")) p2 <- ggplot(data %>% filter(!celltype %in% c("Erythrocyte","undefined")),# aes(UMAP_1,UMAP_2,color=celltype))+ geom_point(size=1)+ facet_wrap(~group)+ ggthemes::theme_few()+ theme_test(12)+ #scale_size_continuous("-log10(Fisher's P)",range = c(1,4))+ theme(#legend.position = "none", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9),#,angle = 90,vjust = 0.5,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), #axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ ggsci::scale_color_npg()+ guides(color = guide_legend(override.aes = list(size=5))) library(patchwork) p <- p1/p2 ggsave(p,filename = "../../Hetero.GSE210347.BCAA.AUCell.Score.Colorectal.pdf", height = 7.5,width = 10) ######## BCAT1 AML ######## setwd("K:/TCGA/Anlysis/BCAA/BCAT1.AML") library(tidyverse) library(GSVA) library(GSEABase) library(clusterProfiler) library(org.Hs.eg.db) library(enrichplot) library(limma) library(tidyverse) library(TCGAbiolinks) library(DESeq2) library(fgsea) GO.KEGG.Enrichment <- function(Symbol,filename){ GO.Enrich <- enrichGO( gene = Symbol, keyType="SYMBOL", OrgDb = "org.Hs.eg.db", ont = "ALL", pAdjustMethod = "BH", qvalueCutoff = 0.05 ) if (length(GO.Enrich) != 0) { write.csv(GO.Enrich@result,file = paste(filename,".Enrichment.GO.",length(Symbol),".csv",sep = "")) print("GO Done") } gene.id <- bitr(Symbol,fromType = 'SYMBOL',toType = c('ENTREZID'),OrgDb="org.Hs.eg.db") KEGG.Enrich <- enrichKEGG(gene = na.omit(gene.id$ENTREZID), organism = "hsa", keyType = "kegg", pvalueCutoff = 1,qvalueCutoff = 1, pAdjustMethod = "BH", minGSSize = 5,use_internal_data = FALSE) if (length(KEGG.Enrich) != 0) { KEGG.Enrich2<-setReadable(KEGG.Enrich, OrgDb = org.Hs.eg.db, keyType="ENTREZID") write.csv(KEGG.Enrich2@result,file = paste(filename,".Enrichment.KEGG.",length(na.omit(gene.id$ENTREZID)),".csv",sep = "")) print("KEGG Done") } #GO.KEGG=list() #GO.KEGG[["GO"]] = GO.Enrich #GO.KEGG[["KEGG"]] = KEGG.Enrich2 #return(GO.KEGG) } Hallmark <- read.gmt("../../../msigdb/h.all.v7.5.1.symbols.gmt") KEGG <- read.gmt("../../../msigdb/c2.cp.kegg.v7.5.1.symbols.gmt") GO <- read.gmt("../../../msigdb/c5.go.v7.5.1.symbols.gmt") #### GSE100778 => #### setwd("K:/TCGA/Anlysis/BCAA/BCAT1.AML") data1 <- read.csv("GSE100778.SKM1.top.table.tsv",sep = "\t") # BCAT1 Down data1 <- data1 %>% filter(Gene.symbol != "") ge = data1$logFC names(ge) = data1$Gene.symbol ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.SKM1.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.SKM1.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.SKM1.GSEA.GO.RDS") data1 <- read.csv("GSE100778.MOLM13.top.table.tsv",sep = "\t") # BCAT1 Down data1 <- data1 %>% filter(Gene.symbol != "") ge = data1$logFC names(ge) = data1$Gene.symbol ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.MOLM13.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.MOLM13.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.MOLM13.GSEA.GO.RDS") data1 <- read.csv("GSE100778.HL60.top.table.tsv",sep = "\t") # BCAT1 Down data1 <- data1 %>% filter(Gene.symbol != "") ge = data1$logFC names(ge) = data1$Gene.symbol ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.HL60.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.HL60.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE100778.HL60.GSEA.GO.RDS") HALLMARK.R1 <- readRDS("GSE100778.SKM1.GSEA.HALLMARK.RDS") HALLMARK.R2 <- readRDS("GSE100778.MOLM13.GSEA.HALLMARK.RDS") HALLMARK.R3 <- readRDS("GSE100778.HL60.GSEA.HALLMARK.RDS") TERMS <- c("HALLMARK_MYC_TARGETS_V1", "HALLMARK_MYC_TARGETS_V2", "HALLMARK_P53_PATHWAY", "HALLMARK_G2M_CHECKPOINT", "HALLMARK_E2F_TARGETS", "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION", "HALLMARK_MTORC1_SIGNALING", "HALLMARK_APOPTOSIS", "HALLMARK_FATTY_ACID_METABOLISM") HALLMARK.D1 <- HALLMARK.R1@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="SKM1") %>% mutate(NES=-NES) HALLMARK.D2 <- HALLMARK.R2@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="MOLM13")%>% mutate(NES=-NES) HALLMARK.D3 <- HALLMARK.R2@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="HL60")%>% mutate(NES=-NES) HALLMARK.D <- rbind.data.frame(HALLMARK.D1,HALLMARK.D2) %>% rbind.data.frame(HALLMARK.D3) library(RColorBrewer) p1 <- ggplot(HALLMARK.D,aes(CellLine,ID))+ #geom_segment(aes(x=0,xend=,y=ID,yend=ID),color="black")+ geom_point(aes(fill=NES,size=-log10(p.adjust)),shape=21,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title="")+ theme(plot.title = element_text(vjust = 0.5))+ scale_fill_gradient2(high = brewer.pal(11,"RdBu")[1], low = brewer.pal(11,"RdBu")[11], mid = "white")+ #,breaks=c(1.4,1.8,2.2) #ggsci::scale_fill_npg()+ #scale_x_continuous(limits = c(0,3))+ scale_size_continuous(range = c(2,4)) #,breaks = c(2,5,8) ggsave(p1,filename = "GSE100778.GSEA.HALLMARK.pdf",height = 3,width = 4.3) KEGG.R1 <- readRDS("GSE100778.SKM1.GSEA.KEGG.RDS") KEGG.R2 <- readRDS("GSE100778.MOLM13.GSEA.KEGG.RDS") KEGG.R3 <- readRDS("GSE100778.HL60.GSEA.KEGG.RDS") TERMS <- c("KEGG_APOPTOSIS", "KEGG_DNA_REPLICATION", "KEGG_CELL_CYCLE", "KEGG_P53_SIGNALING_PATHWAY", "KEGG_FATTY_ACID_METABOLISM", "KEGG_FOCAL_ADHESION", "KEGG_HOMOLOGOUS_RECOMBINATION", "KEGG_MISMATCH_REPAIR", "KEGG_BASE_EXCISION_REPAIR", "KEGG_MTOR_SIGNALING_PATHWAY") KEGG.D1 <- KEGG.R1@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"KEGG_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="SKM1") %>% mutate(NES=-NES) KEGG.D2 <- KEGG.R2@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"KEGG_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="MOLM13")%>% mutate(NES=-NES) KEGG.D3 <- KEGG.R2@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"KEGG_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="HL60")%>% mutate(NES=-NES) KEGG.D <- rbind.data.frame(KEGG.D1,KEGG.D2) %>% rbind.data.frame(KEGG.D3) p1 <- ggplot(KEGG.D,aes(CellLine,ID))+ #geom_segment(aes(x=0,xend=,y=ID,yend=ID),color="black")+ geom_point(aes(fill=NES,size=-log10(p.adjust)),shape=21,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title="")+ theme(plot.title = element_text(vjust = 0.5))+ scale_fill_gradient2(high = brewer.pal(11,"RdBu")[1], low = brewer.pal(11,"RdBu")[11], mid = "white")+ #,breaks=c(1.4,1.8,2.2) #ggsci::scale_fill_npg()+ #scale_x_continuous(limits = c(0,3))+ scale_size_continuous(range = c(2,4),breaks = c(2,4,6)) #,breaks = c(2,5,8) ggsave(p1,filename = "GSE100778.GSEA.KEGG.pdf",height = 3,width = 4.6) library(GseaVis) gseaNb(object = KEGG.R2, geneSetID = "KEGG_CELL_CYCLE", addPval = T, pvalX = 0.6,pvalY = 0.75, pCol = 'black', pHjust = 0,subPlot = 2) #### GSE100779-82 => Methylation => Linux #### files <- list.files(".",pattern="tsv$") files <- files[str_detect(files,"Methylation")] #### GSE103960 shBCAT1 #### data1 <- read.csv("GSE103960.HD48.shBCAT1.top.table.tsv",sep = "\t") # Ctrl vs KD data1 <- data1 %>% filter(Gene.symbol != "") ge = -data1$logFC # BCAT1 Down names(ge) = data1$Gene.symbol ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE103960.HD48.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE103960.HD48.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE103960.HD48.GSEA.GO.RDS") data1 <- read.csv("GSE103960.AML131.shBCAT1.top.table.tsv",sep = "\t") # Ctrl vs KD data1 <- data1 %>% filter(Gene.symbol != "") ge = -data1$logFC # BCAT1 Down names(ge) = data1$Gene.symbol ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE103960.AML131.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE103960.AML131.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE103960.AML131.GSEA.GO.RDS") HALLMARK.R1 <- readRDS("GSE103960.AML131.GSEA.HALLMARK.RDS") HALLMARK.R2 <- readRDS("GSE103960.HD48.GSEA.HALLMARK.RDS") TERMS <- c("HALLMARK_MYC_TARGETS_V1", "HALLMARK_G2M_CHECKPOINT", "HALLMARK_E2F_TARGETS", "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION", "HALLMARK_MTORC1_SIGNALING", "HALLMARK_APOPTOSIS", "HALLMARK_FATTY_ACID_METABOLISM") library(GseaVis) HALLMARK.D1 <- HALLMARK.R1@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="AML131") HALLMARK.D2 <- HALLMARK.R2@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="HD48") HALLMARK.D <- rbind.data.frame(HALLMARK.D1,HALLMARK.D2) library(RColorBrewer) p1 <- ggplot(HALLMARK.D,aes(CellLine,ID))+ #geom_segment(aes(x=0,xend=,y=ID,yend=ID),color="black")+ geom_point(aes(fill=NES,size=-log10(p.adjust)),shape=21,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title="")+ theme(plot.title = element_text(vjust = 0.5))+ scale_fill_gradientn(colours = brewer.pal(9,"Reds"),breaks=c(1.4,1.8,2.2))+ #ggsci::scale_fill_npg()+ #scale_x_continuous(limits = c(0,3))+ scale_size_continuous(range = c(2,4),breaks = c(2,5,8)) ggsave(p1,filename = "GSE103960.AML131.GSEA.HALLMARK.pdf",height = 3,width = 5) KEGG.R1 <- readRDS("GSE103960.AML131.GSEA.KEGG.RDS") KEGG.R2 <- readRDS("GSE103960.HD48.GSEA.KEGG.RDS") TERMS <- c("KEGG_APOPTOSIS", "KEGG_DNA_REPLICATION", "KEGG_CELL_CYCLE", "KEGG_P53_SIGNALING_PATHWAY", "KEGG_FATTY_ACID_METABOLISM", "KEGG_FOCAL_ADHESION", "KEGG_HOMOLOGOUS_RECOMBINATION", "KEGG_MISMATCH_REPAIR", "KEGG_BASE_EXCISION_REPAIR", "KEGG_MTOR_SIGNALING_PATHWAY") library(GseaVis) KEGG.D1 <- KEGG.R1@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"KEGG_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="AML131") KEGG.D2 <- KEGG.R2@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"KEGG_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) %>% mutate(CellLine="HD48") KEGG.D <- rbind.data.frame(KEGG.D1,KEGG.D2) library(RColorBrewer) p1 <- ggplot(KEGG.D,aes(CellLine,ID))+ #geom_segment(aes(x=0,xend=,y=ID,yend=ID),color="black")+ geom_point(aes(fill=NES,size=-log10(p.adjust)),shape=21,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="",y="",title="")+ theme(plot.title = element_text(vjust = 0.5))+ scale_fill_gradientn(colours = brewer.pal(9,"Reds"),breaks=c(1.6,2,2.4))+ #ggsci::scale_fill_npg()+ #scale_x_continuous(limits = c(0,3))+ scale_size_continuous(range = c(2,4),breaks = c(2,6,10)) ggsave(p1,filename = "GSE103960.AML131.GSEA.KEGG.pdf",height = 3,width = 4.3) #### GSE127181 => gabapentin => BCAT1 inhibitor #### GPL <- read.csv("GPL21185-21174.txt",sep = "\t",header = T) data1 <- read.csv("GSE127181.top.table.tsv",sep = "\t") %>% merge(GPL,by="ID") %>% filter(GENE_SYMBOL != "") ge = data1$logFC # names(ge) = data1$GENE_SYMBOL ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE127181.AML.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE127181.AML.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE127181.AML.GSEA.GO.RDS") HALLMARK.R <- readRDS("GSE127181.AML.GSEA.HALLMARK.RDS") TERMS <- c("HALLMARK_MYC_TARGETS_V1", "HALLMARK_MYC_TARGETS_V2", "HALLMARK_G2M_CHECKPOINT", "HALLMARK_E2F_TARGETS", "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION", "HALLMARK_MTORC1_SIGNALING", "HALLMARK_APOPTOSIS", "HALLMARK_FATTY_ACID_METABOLISM") library(GseaVis) HALLMARK.D <- HALLMARK.R@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"HALLMARK_")) %>% mutate(ID=factor(ID,levels = rev(.$ID))) p1 <- ggplot(HALLMARK.D,aes(NES,ID))+ geom_segment(aes(x=0,xend=NES,y=ID,yend=ID),color="black")+ geom_point(aes(fill=ID,size=-log10(p.adjust)),shape=21,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="NES",y="",title="Ctrl vs Gabapentin")+ theme(plot.title = element_text(vjust = 0.5))+ ggsci::scale_fill_npg()+ scale_x_continuous(limits = c(0,3))+ scale_size_continuous(range = c(2,4)) ggsave(p1,filename = "GSE127181.AML.GSEA.HALLMARK.pdf",height = 3,width = 8) KEGG.R <- readRDS("GSE127181.AML.GSEA.KEGG.RDS") TERMS <- c("KEGG_APOPTOSIS", "KEGG_DNA_REPLICATION", "KEGG_FATTY_ACID_METABOLISM", "KEGG_FOCAL_ADHESION", "KEGG_HOMOLOGOUS_RECOMBINATION", "KEGG_MISMATCH_REPAIR", "KEGG_BASE_EXCISION_REPAIR", "KEGG_MTOR_SIGNALING_PATHWAY") KEGG.D <- KEGG.R@result %>% filter(ID %in% TERMS) %>% mutate(ID=str_remove_all(ID,"KEGG_")) %>% arrange(desc(NES)) %>% mutate(ID=factor(ID,levels = rev(.$ID))) p2 <- ggplot(KEGG.D,aes(NES,ID))+ geom_segment(aes(x=0,xend=NES,y=ID,yend=ID),color="black")+ geom_point(aes(fill=ID,size=-log10(p.adjust)),shape=21,color="black")+ ggthemes::theme_few()+ theme(#legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(x="NES",y="",title="Ctrl vs Gabapentin")+ theme(plot.title = element_text(vjust = 0.5))+ ggsci::scale_fill_npg()+ scale_x_continuous(limits = c(0,3))+ scale_size_continuous(range = c(2,4)) ggsave(p2,filename = "GSE127181.AML.GSEA.KEGG.pdf",height = 3,width = 7) GO.R <- readRDS("GSE127181.AML.GSEA.GO.RDS") #### GSE148701 => Macrophage #### library(limma) library(data.table) files <- list.files("GSE148701_RAW/",pattern = "gz$") data <- data.frame() for (file in files) { data <- fread(file.path("GSE148701_RAW/",file)) %>% #t() %>% data.frame(check.names =F ) %>% remove_rownames() %>% column_to_rownames("V1") %>% t() %>% rbind.data.frame(data) } data <- data %>% t() %>% data.frame(check.names = F) %>% magrittr::set_colnames(c("Ctrl1","KD1","Ctrl2","KD2","Ctrl3","KD3")) data <- data %>% magrittr::set_colnames(rev(c("Ctrl1","KD1","Ctrl2","KD2","Ctrl3","KD3"))) write.csv(data,file = file.path("GSE148701_RAW/","Count.csv")) ### DEG library(DESeq2) countData <- read.csv(file.path("GSE148701_RAW/","Count.csv"),row.names = 1) countData <- countData[rowMeans(countData)>1,] condition <- factor(c(rep(c("Ctrl","KD"),length=6))) colData <- data.frame(row.names=colnames(countData), condition) dds <- DESeqDataSetFromMatrix(countData = countData, colData = colData, design = ~ condition) dds1 <- DESeq(dds) res <- results(dds1) res1 <- data.frame(res, stringsAsFactors = FALSE, check.names = FALSE) res1 <- res1[order(res1$pvalue, res1$log2FoldChange, decreasing = c(FALSE, TRUE)), ] write.csv(res1,file = "GSE148701.Macrophage.BCAT1.KD.csv",row.names = F) GeneType <- read.table("../../../../../Metastasis/TissueMicrobiome/Human.GRC38.GeneType.txt",header = T,sep = ",") GeneType.Unique <- GeneType %>% dplyr::select(Gene.stable.ID,Gene.name,NCBI.gene..formerly.Entrezgene..ID) %>% unique() %>% magrittr::set_colnames(c("ENSEMBL","SYMBOL","EntreZ")) countData2 <- countData %>% mutate(ENSEMBL=str_remove_all(rownames(.),"\\..*")) Res <- res1 %>% mutate(ENSEMBL=str_remove_all(rownames(.),"\\..*")) %>% merge(GeneType.Unique,by="ENSEMBL") %>% merge(countData2,by="ENSEMBL") %>% arrange(pvalue) %>% mutate(padj=p.adjust(pvalue,method = "BH",n = nrow(.))) %>% filter(SYMBOL != "") write.csv(Res,file = "GSE148701.Macrophage.BCAT1.KD.csv",row.names = F) ge = Res$log2FoldChange # names(ge) = Res$SYMBOL ge = sort(ge,decreasing = T) GSEA.Res <- GSEA(ge, TERM2GENE = Hallmark,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE148701.Macrophage.GSEA.HALLMARK.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = KEGG,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE148701.Macrophage.GSEA.KEGG.RDS") GSEA.Res <- GSEA(ge, TERM2GENE = GO,eps = 1e-100,pvalueCutoff =0.05) saveRDS(GSEA.Res,file = "GSE148701.Macrophage.GSEA.GO.RDS") HALLMARK.R <- readRDS("GSE148701.Macrophage.GSEA.HALLMARK.RDS") library(GseaVis) gseaNb(object = HALLMARK.R, geneSetID = "HALLMARK_REACTIVE_OXYGEN_SPECIES_PATHWAY", addPval = T, pvalX = 0.6,pvalY = 0.75, pCol = 'black', pHjust = 0,subPlot = 2) KEGG.R <- readRDS("GSE148701.Macrophage.GSEA.KEGG.RDS") GO.R <- readRDS("GSE148701.Macrophage.GSEA.GO.RDS") Res <- read.csv("GSE148701.Macrophage.BCAT1.KD.csv") Up.BCAT1.KD <- Res %>% filter(log2FoldChange <= -1) %>% filter(padj <= 0.05) GO.KEGG.Enrichment(Up.BCAT1.KD$SYMBOL,"GSE148701.Macrophage.Up.BCAT1.KD") Up.BCAT1.Ctrl <- Res %>% filter(log2FoldChange >= 1) %>% filter(padj <= 0.05) GO.KEGG.Enrichment(Up.BCAT1.Ctrl$SYMBOL,"GSE148701.Macrophage.Up.BCAT1.Ctrl") #### Methylation => EPIC => ChAMP #### library("ChAMP") myLoad <- champ.load("GSE100779",arraytype="EPIC") save(myLoad,file = "GSE100779.ChAMP.load.Rda") #CpG.GUI(arraytype="EPIC") champ.QC() # Alternatively QC.GUI(arraytype="EPIC") myNorm <- champ.norm(arraytype="EPIC") save(myNorm,file = "GSE100779.ChAMP.Norm.Rda") #champ.SVD() # If Batch detected, run champ.runCombat() here.This data is not suitable. #myDMP <- champ.DMP(beta = myNorm,pheno=myLoad$pd$Sample_Group,arraytype="EPIC") #DMP.GUI(DMP=myDMP[[1]],beta=myNorm,pheno=myLoad$pd$Sample_Group,arraytype="EPIC") #save(myDMP,file="GSE100779.DMP.rda") myDMR <- champ.DMR(beta=myNorm, pheno=myLoad$pd$Sample_Group, compare.group=c("Ctrl", "BCAT1OE"), arraytype="EPIC",method="Bumphunter") save(myDMR,file="GSE100779.DMR.rda") DMR.GUI(arraytype="EPIC") myGSEA <- champ.GSEA(arraytype="EPIC") save(myGSEA,file="GSE100779.GSEA.rda") #### GSE100782 => Methylation #### setwd("K:/TCGA/Anlysis/BCAA/BCAT1.AML") library(tidyverse) library(data.table) Methy.Data <- read.csv("GSE100782.MOLM13.Methylation.OE.20WEEK.top.table.tsv",sep = "\t") %>% filter(adj.P.Val<=0.05) #%>% filter(abs(logFC)>=0.5) Anno <- read.csv("GPL21145_MethylationEPIC_15073387_v-1-0.csv",skip = 7,header = T) %>% dplyr::select(IlmnID,Name,UCSC_RefGene_Name,UCSC_RefGene_Accession,UCSC_RefGene_Group,UCSC_CpG_Islands_Name,Relation_to_UCSC_CpG_Island,Phantom4_Enhancers,Phantom5_Enhancers,DMR,HMM_Island,Regulatory_Feature_Name,Regulatory_Feature_Group) %>% filter(UCSC_RefGene_Name != "") Des <- Anno %>% filter(Name %in% Methy.Data$ID) %>% mutate(Gene.symbol=str_remove_all(UCSC_RefGene_Name,"\\;.*")) %>% filter(Gene.symbol %in% Gene.SYMBOL) %>% merge(Methy.Data,.,by.x="ID",by.y="Name") %>% filter(!str_detect(UCSC_RefGene_Group,"Body")) %>% arrange(Gene.symbol) %>% mutate(ID=factor(ID,levels = .$ID)) p1 <- ggplot(Des,aes(ID,logFC))+ geom_segment(aes(x=ID,xend=ID,y=0,yend=logFC),color="black")+ geom_point(aes(fill=Gene.symbol,size=-log10(adj.P.Val)),shape=21,color="black")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_text(size=9,angle = 45,vjust = 0.5,hjust = 1),#element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), #axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(y="logFC",x="",title="Ctrl vs BCAT1 OE")+ theme(plot.title = element_text(hjust = 0.5))+ ggsci::scale_fill_npg()+ scale_y_continuous(limits = c(-0.2,0))+ geom_hline(yintercept = c(-0.05,-0.1,-0.15),linetype=2)+ scale_size_continuous(range = c(2,4)) p2 <- ggplot(Des %>% dplyr::select(ID,Gene.symbol) %>% unique(), aes(ID,1,fill=Gene.symbol))+ geom_col(width = 1)+ theme_void()+ ggsci::scale_fill_npg() p12 <- p1 %>% aplot::insert_bottom(p2,height = 0.2) ggsave(p12,filename = "SYMBOL.EXAMPLE.Methylation.pdf",height = 2,width = 9) ggsave(p12,filename = "SYMBOL.EXAMPLE.Methylation.Legend.pdf",height = 8,width = 9) write.csv(Des,file = "SYMBOL.EXAMPLE.Methylation.Figure.Data.csv",row.names = F) data1 <- read.csv("GSE100778.SKM1.top.table.tsv",sep = "\t") %>% filter(adj.P.Val<=0.05) %>% filter(Gene.symbol %in%Gene.SYMBOL) %>% mutate(Project="GSE100778.SKM1") data2 <- read.csv("GSE100778.MOLM13.top.table.tsv",sep = "\t")%>% filter(adj.P.Val<=0.05)%>% filter(Gene.symbol %in%Gene.SYMBOL) %>% mutate(Project="GSE100778.MOLM13") data3 <- read.csv("GSE100778.HL60.top.table.tsv",sep = "\t")%>% filter(adj.P.Val<=0.05)%>% filter(Gene.symbol %in%Gene.SYMBOL) %>% mutate(Project="GSE100778.HL60") data4 <- read.csv("GSE103960.HD48.shBCAT1.top.table.tsv",sep = "\t")%>% filter(adj.P.Val<=0.05) %>% filter(Gene.symbol %in%Gene.SYMBOL) %>% mutate(logFC=-logFC)%>% mutate(Project="GSE103960.HD48")# Ctrl vs KD data <- rbind.data.frame(data1,data2) %>% rbind.data.frame(data3) %>% rbind.data.frame(data4) Figure.Data <- data %>% filter(Gene.symbol != "BCAT1") %>% mutate(CellLine=str_remove_all(Project,".*\\.")) %>% mutate(Cancer=str_remove_all(Project,"\\..*")) %>% arrange(Gene.symbol) %>% mutate(Xaxis=paste(Project,ID,sep = "_")) %>% mutate(Xaxis=factor(Xaxis,levels = .$Xaxis)) p1 <- ggplot(Figure.Data,aes(Xaxis,logFC))+ geom_segment(aes(x=Xaxis,xend=Xaxis,y=0,yend=logFC),color="black")+ geom_point(aes(fill=Gene.symbol,size=-log10(adj.P.Val)),shape=21,color="black")+ ggthemes::theme_few()+ theme(legend.position = "bottom", axis.text = element_text(color = "black"), axis.text.x = element_blank(),#element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #element_text(size=9,angle = 90,vjust = 0.3,hjust = 1), #,angle = 90,vjust = 0.3,hjust = 1 axis.text.y = element_text(size=9), axis.ticks.x = element_blank(), axis.title = element_text(size=12), panel.border = element_rect(fill=NA,color="black", size=1.2, linetype="solid"))+ labs(y="logFC",x="",title="Ctrl vs shBCAT1")+ theme(plot.title = element_text(hjust = 0.5))+ ggsci::scale_fill_npg()+ scale_y_continuous(limits = c(-1.7,0))+ geom_hline(yintercept = c(-0.5,-1,-1.5),linetype=2)+ scale_size_continuous(range = c(2,4)) p2 <- ggplot(Figure.Data %>% dplyr::select(Xaxis,Gene.symbol) %>% unique(), aes(Xaxis,1,fill=Gene.symbol))+ geom_col(width = 1)+ theme_void()+ ggsci::scale_fill_npg() library(RColorBrewer) p3 <- ggplot(Figure.Data %>% dplyr::select(Xaxis,Cancer) %>% unique(), aes(Xaxis,1,fill=Cancer))+ geom_col(width = 1)+ theme_void()+ scale_fill_manual("",values = brewer.pal(8,"Paired")) p123 <- p1 %>% aplot::insert_bottom(p2,height = 0.2) %>% aplot::insert_bottom(p3,height = 0.1) ggsave(p123,filename = "SYMBOL.EXAMPLE.pdf",height = 2,width = 10) ggsave(p123,filename = "SYMBOL.EXAMPLE.Legend.pdf",height = 8,width = 10) write.csv(Figure.Data,file = "SYMBOL.EXAMPLE.Expression.Figure.Data.csv",row.names = F) #GPL <- read.csv("GPL21185-21174.txt",sep = "\t",header = T) #data6 <- read.csv("GSE127181.top.table.tsv",sep = "\t") %>% # merge(GPL,by="ID") %>% filter(GENE_SYMBOL != "") %>% # filter(adj.P.Val<=0.05) #data5 <- read.csv("GSE103960.AML131.shBCAT1.top.table.tsv",sep = "\t")%>% # filter(adj.P.Val<=0.05) Gene.SYMBOL <- intersect(data1$Gene.symbol,data2$Gene.symbol) %>% intersect(data3$Gene.symbol) %>% intersect(data4$Gene.symbol) %>% #intersect(data6$GENE_NAME) %>% intersect(Des$Gene.symbol) %>% sort() ######## Cancer cell States Signature #### files <- list.files("../../CancerSEA",pattern = ".txt$") #data <- read.csv("../../mRNA.PanCancer.Exp/TCGA-LIHC.mRNA.TPM.csv",check.names = F,row.names = 1) library(data.table) Score.Gene <- list() for (file in files) { SigName = file %>% str_remove_all("\\..*") data2 <- fread(file.path("../../CancerSEA",file)) Score.Gene[[SigName]] = data2$EnsembleID } files <- list.files(path = "../../mRNA.PanCancer.Exp/",pattern = "csv$") Pancancer.CancerState <- data.frame() for (file in files) { project <- file %>% str_remove_all("\\..*") %>% str_remove_all("TCGA-") print(project) phenotype <- read.csv(paste("../../Clinical.XENA/",project,"/","TCGA-",project,".GDC_phenotype.tsv",sep = ""),sep = "\t",header = T) %>% dplyr::filter(sample_type.samples == "Primary Tumor" | sample_type.samples == "Primary Blood Derived Cancer - Peripheral Blood") TPM <- read.csv(paste("../../mRNA.PanCancer.Exp/TCGA-",project,".mRNA.TPM.csv",sep = ""),check.names = F,row.names = 1) TPM.log <- log2(TPM+1) GSEA.Res <- gsva(expr=as.matrix(TPM.log), gset.idx.list=Score.Gene, method="ssgsea", kcdf="Gaussian" , verbose=T) Pancancer.CancerState <- GSEA.Res %>% t() %>% data.frame(check.names = F) %>% mutate(Cancer=project) %>% rbind.data.frame(Pancancer.CancerState) } write.csv(Pancancer.CancerState,"../../CancerSEA/Pancancer.CancerState.csv") #### RS-high vs RS-low => Stage #### RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% mutate(ID=substring(SampleID,1,15)) PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") Risk.Cutoff <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.CutOff.csv") Stage.Clinical <- read_xlsx("../../PanCancer.DiverseIndex.xlsx",sheet = 19) data2<-Stage.Clinical[,-1] # 18 Grade 19 Stage Index.data2 <- merge(RiskScore.Cancer,data2,by="ID",by.y="SampleName") %>% unique() %>% data.frame() %>% filter(SampleID %in% PhenoType$Sample.ID) %>% mutate(Stage=str_remove_all(Stage,"Stage ")) %>% #gather(Attribute,Score,StromalScore:ESTIMATEScore) %>% #dplyr::select(ID,SampleID,Attribute,Score,Cancer,total_risk_score) %>% group_by(Cancer) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) write.csv(Index.data2,file = "RiskScore.Stage.csv",row.names = F) Fisher.Stage <- data.frame() for(cancer in unique(Index.data2$Cancer)){ mid.data <- Index.data2 %>% filter(Cancer==cancer) table <- table(mid.data$Group,mid.data$Stage) test <- fisher.test(table) Fisher.Stage <- data.frame(Fisher.P = test$p.value,Cancer=cancer) %>% rbind.data.frame(Fisher.Stage) } write.csv(Fisher.Stage,file = "RiskScore.Stage.Fisher.P.csv") col <- brewer.pal(9,"Greens")[c(2,4,5,6)] #c("#A6CEE3","#FDBF6F","#CAB2D6","#E31A1C") names(col) <- sort(unique(Index.data2$Stage)) Figure.plot <- list() for (cancer in unique(Index.data2$Cancer)) { Index.dat3 <- Index.data2 %>% filter(Cancer==cancer) %>% group_by(Cancer,Group,Stage) %>% summarise(Count=n()) %>% spread(Group,Count,fill=0) dat1 = aggregate(Index.dat3$High, by = list(Index.dat3$Stage), FUN = sum) dat1$per1 = dat1$x / sum(dat1$x) # for循环构建标签的相对位置 for (i in seq(nrow(dat1), 1)) { if (i == nrow(dat1)) { dat1$per.y1[i] = dat1$per1[i] / 2 }else{ dat1$per.y1[i] = sum(dat1$per1[(i + 1):nrow(dat1)]) + dat1$per1[i] / 2 } } dat1$label1 = paste(dat1$Group.1,'(',round(dat1$per1*100, 2),'%',')', sep = '') Index.dat3 = merge(Index.dat3, dat1[,c(1,3,4,5)], by.x = 'Stage', by.y = 'Group.1') dat2 = aggregate(Index.dat3$Low, by = list(Index.dat3$Stage), FUN = sum) dat2$per2 = dat2$x / sum(dat2$x) for (i in seq(nrow(dat2), 1)) { if (i == nrow(dat2)) { dat2$per.y2[i] = dat2$per2[i] / 2 }else{ dat2$per.y2[i] = sum(dat2$per2[(i + 1):nrow(dat2)]) + dat2$per2[i] / 2 } } dat2$label2 = paste(dat2$Group.1,'(',round(dat2$per2*100, 2),'%',')', sep = '') Index.dat3 = merge(Index.dat3, dat2[,c(1,3,4,5)], by.x = 'Stage', by.y = 'Group.1') pvalue <- Fisher.Stage %>% filter(Cancer==cancer) p <- ggplot(Index.dat3) + # 绘制柱状图 geom_bar(aes("a", per2, fill = Stage), stat = 'identity', width = 1.3) + # 添加标签 #geom_text(aes(1.25, as.numeric(per.y2), # label = label2), # size =2.5, color = 'black') + # 绘制柱状图 geom_bar(aes("b", per1, fill = Stage), stat = 'identity', width = .8, color = 'white') + # 添加标签 #geom_text(aes(2, as.numeric(per.y1),label = label1), # size = 2.5, color = 'black') + # 设置Y轴刻度 scale_y_continuous(labels = scales::percent) + coord_polar(theta = "y") + # 转换坐标轴 theme_void() + scale_fill_manual(values = col)+ #ggsci::scale_fill_npg() + # 设置填充色 theme(legend.position = 'none')+ # 隐藏图例 ggtitle(paste(cancer,"\n","Fisher'P:",round(pvalue$Fisher.P,4),sep = ""))+ theme(plot.title = element_text(hjust = 0.5)) ggsave(p,filename = paste("RS.Stage.",cancer,".PiePlot.pdf",sep = ""), height = 2.5,width = 2.5) Figure.plot[[cancer]] <- p } p321<-cowplot::plot_grid(plotlist = Figure.plot, ncol = 4,align = 'hv') ggsave(p321,filename = "RS.Stage.PanCancer.Fisher.pdf",height = 10,width = 8) #### RS-high vs RS-low => Cancer State #### Pancancer.CancerState <- read.csv("../../CancerSEA/Pancancer.CancerState.csv") PhenoType <- read.csv("../../mRNA.Phenotype.csv") PhenoType <- PhenoType %>% filter(Sample.Type == "Primary Tumor" | Sample.Type == "Solid Tissue Normal" | Sample.Type == "Primary Blood Derived Cancer - Peripheral Blood") %>% dplyr::select(Project.ID,Sample.ID,Sample.Type) %>% mutate(Type=ifelse(Sample.Type=="Solid Tissue Normal","Normal","Tumor")) %>% mutate(Cancer=str_remove(Project.ID,"TCGA-")) %>% filter(Type=="Tumor") setwd("K:/TCGA/Anlysis/BCAA") RiskScore.Cancer <- read.csv(file = "RiskGroup.COX.TCGA.RiskScore.csv") %>% #mutate(ID=substring(SampleID,1,15))%>% unique() %>% data.frame() %>% filter(SampleID %in% PhenoType$Sample.ID) %>% group_by(Cancer) %>% mutate(cutpoint=median(total_risk_score)) %>% mutate(Group = if_else(total_risk_score >= cutpoint,"High","Low")) %>% merge(Pancancer.CancerState %>% dplyr::select(-Cancer),by.x="SampleID",by.y="X") %>% dplyr::select(-cutpoint,-total_risk_score,-OS,-OS.time) %>% arrange(Cancer,Group) Test.State <- RiskScore.Cancer %>% gather(State,Score,Angiogenesis:Quiescence) %>% group_by(Cancer,State) %>% rstatix::t_test(Score~Group,detailed = T) write.csv(Test.State,"RiskScore.CancerState.Ttest.csv") rt <- data.frame() for(cancer in unique(RiskScore.Cancer$Cancer)){ rt <- RiskScore.Cancer %>% filter(Cancer==cancer) %>% dplyr::select(-Cancer,-Group) %>% remove_rownames() %>% column_to_rownames("SampleID") %>% scale() %>% rbind(rt) } library(ComplexHeatmap) library(circlize) library(ggsci) color = c(pal_d3("category20")(20),pal_d3("category20b")(20),pal_d3("category20c")(20),pal_d3("category10")(10)) col1 <- color[1:23] names(col1) <- sort(unique(RiskScore.Cancer$Cancer)) col2 <- c("Black","Red") names(col2) <- c("High","Low") ha = HeatmapAnnotation(Cancer= RiskScore.Cancer$Cancer, RS = RiskScore.Cancer$Group, show_legend = rep(TRUE, 2),#是否要显示annotation legend annotation_height = unit(rep(5,2), "mm"),#临床annotation的高度 col = list(Cancer=col1, RS=col2), annotation_legend_param = list( Cancer = list(title = "Cancers"), RS = list(title = "RS")), simple_anno_size = unit(0.3, "cm"), na_col = "white", annotation_name_side = "left" ) ht_opt(legend_border='black', heatmap_border = TRUE, annotation_border = TRUE) col_fun <- colorRamp2(c(-2, 0,2), c("#377EB8", "white", "#E41A1C")) subtype <- RiskScore.Cancer$Cancer ht <- Heatmap(t(rt), col = col_fun, name = "Z Score", cluster_rows = F, cluster_columns = T, show_row_names = TRUE, row_names_side = "left", #行名位置 show_column_names = FALSE, column_split = subtype, #根据数据修改 #row_split = c(rep(1,3),rep(2,6),rep(3,15),rep(4,1),rep(5,6),rep(6,3),rep(7,6)),#根据数据修改 #row_gap = unit(c(1,1,1,1,1,1) ,"mm"), column_gap = unit(c(2), "mm"), row_title = NULL,column_title = NULL, #right_annotation = harow, use_raster=F, top_annotation = ha) pdf("RiskScore.CancerState.pdf",width = 16,height = 6) draw(ht,padding = unit(c(0.1, 4, 0.1, 0.5), "cm"), annotation_legend_side = "right", heatmap_legend_side = "right") dev.off() #################### #################### #################### library(TCGAbiolinks) projects <- getGDCprojects() projects <- projects$project_id TCGA_dowload<-function(x,dirpath){ query.rppa <- GDCquery( project = x, data.category = "Proteome Profiling", data.type = "Protein Expression Quantification" ) GDCdownload(query.rppa) Proteins <- GDCprepare(query.rppa) saveRDS(Proteins,file = paste0(dirpath,x,"_protein.rds")) } for (j in projects) { print(j) try(TCGA_dowload(j,dirpath = "./CPTAC/"),silent = T) }