dat2 <- dat2[,-14]
dat <- dat2 %>%
mutate(`WID-BC-index` = index,
`immune cell proportion` = ic) %>%
dplyr::select(`WID-BC-index`, `immune cell proportion`, type)
rownames(dat) <- c()
head(dat)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-3e.txt")
# f -------------------------------
ind <- pdat2$typenew %in% c('BRCA2','Control') &
pdat2$sample == 'whole blood'
dat2 <- pdat2[ind,]
type_tmp <- rep('Case',nrow(dat2))
type_tmp[dat2$type=='Control'] <- 'Control'
dat2$type_simple <- factor(type_tmp, levels=c('Case','Control'))
dat2$type <- factor(dat2$type, levels=c('Control','BRCA2'))
dat2 <- dat2[,-14]
dat <- dat2 %>%
mutate(`WID-BC-index` = index,
`immune cell proportion` = ic) %>%
dplyr::select(`WID-BC-index`, `immune cell proportion`, type)
rownames(dat) <- c()
head(dat)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-3f.txt")
#----------------------------------
# Suppl Fig 4
#----------------------------------
# a - c ------------------------------
# cervical internal validation set
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/12-output/pdat_BC.Rdata")
# entire buccal 3c set
load("~/Dropbox/index-dev/3C/1-BC/10-buccal/1-output/pdat.Rdata")
# load buccal phenotype
dat <- read.csv('~/Dropbox/index-dev/3C/0-buccal-preprocess/1-files/BC_RISK_BUCCALS_PHENO_240119.csv')
basename <- dat$buccal_basename
beta_names <- rownames(pdat)
ind <- match(beta_names,basename)
cbasename <- dat$cervical_basename[ind]
ind1 <- match(cbasename, rownames(pdat_BC))
# save subsets of each pdat dataset
pdat_buccal <- pdat[!is.na(ind1),]
pdat_cervix <- pdat_BC[ind[!is.na(ind1)],]
pdat_buccal$type <- plyr::revalue(pdat_buccal$type, c("Breast"="Breast cancer"))
pdat_cervix$type <- plyr::revalue(pdat_cervix$type, c("Breast"="Breast cancer"))
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/12-output/pdat_BC.Rdata")
# entire buccal 3c set
load("~/Dropbox/index-dev/3C/1-BC/10-buccal/1-output/pdat.Rdata")
# load buccal phenotype
dat <- read.csv('~/Dropbox/index-dev/3C/0-buccal-preprocess/1-files/BC_RISK_BUCCALS_PHENO_240119.csv')
ind <- match(rownames(pdat), dat$buccal_basename)
pdat$cervical_base <- dat$cervical_basename[ind]
ind <- match(rownames(pdat_BC), pdat$cervical_base)
pdat <- pdat[ind,]
rm <- !is.na(pdat$cervical_base)
pdat <- pdat[rm,]
pdat_BC <- pdat_BC[rm,]
dat <- pdat_buccal %>%
mutate(`WID-BC-index (epithelial)` = index_epi + index_bm,
`WID-BC-index (immune)` = index_imm,
`WID-BC-index (shared)` = index_shared,
`immune cell proportion` = ic) %>%
dplyr::select(`WID-BC-index (epithelial)`, `WID-BC-index (immune)`, `WID-BC-index (shared)`, `immune cell proportion`, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-4a-c.txt")
# d - f -----------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/16-output/pdat.Rdata")
pdat$type <- plyr::revalue(pdat$type, c("Ovarian"="Ovarian cancer"))
ind <- pdat$dataset=='Training' & pdat$type=='Control'
pdat_OC <- pdat[!ind,]
dat <- pdat_OC %>%
mutate(`WID-BC-index (epithelial)` = index_epi + index_bm,
`WID-BC-index (immune)` = index_imm,
`WID-BC-index (shared)` = index_shared,
`immune cell proportion` = ic) %>%
dplyr::select(`WID-BC-index (epithelial)`, `WID-BC-index (immune)`, `WID-BC-index (shared)`, `immune cell proportion`, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-4e-f.txt")
#----------------------------------
# Suppl Fig 5
#----------------------------------
# a -------------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/1-output/pdat.Rdata")
tdat <- read.csv(file='~/Dropbox/index-dev/3C/0-discovery-preprocess/1-files/PostM_WID_EPIC_LIST_1680_WITH_DATES.csv',sep='\t')
ind2 <- match(rownames(pdat),tdat$Basename)
pdat <- cbind(pdat,
days_rec_to_child=tdat$Days.Rec.to.Child[ind2],
days_col_to_rec=tdat$Days.Col.to.Rec[ind2],
days_col_to_child=tdat$Days.from.Col.to.Child[ind2])
TITLE <- 'Internal Validation'
RANGE <- c(-1.78,2.6)
ind <- !is.na(pdat$days_col_to_child) & (pdat$days_col_to_child<150)
pdat <- pdat[ind,]
dat <- pdat %>%
mutate(days = days_col_to_child,
`WID-BC-index` = index) %>%
dplyr::select(`WID-BC-index`, days, type)
rownames(dat) <- c()
head(dat)
dat %>%
ggplot(aes(x = days,
y = `WID-BC-index`,
colour = type)) +
geom_point() +
geom_smooth(method = "lm")
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-5a.txt")
# b -------------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/1-output/pdat.Rdata")
dat <- read.csv('~/Dropbox/index-dev/3C/0-discovery-preprocess/1-files/PostM_WID_EPIC_LIST_1680_Masterlist_with_phenotypic_data_20180806.csv')
basename <- paste(dat$Sentrix_ID,dat$Sentrix_Position,sep='_')
ind <- match(rownames(pdat), basename)
stnumber <- dat$ST_Number[ind]
#-------------------------------------------#
dat2 <- read.csv(file='~/Dropbox/data/3c/pheno/3C-training-benign-control-subtype-mw-v2.csv')
ind2 <- match(stnumber,dat2$ST_Number)
dat3 <- readxl::read_xlsx('~/Dropbox/data/3c/pheno/PostM_WID_Pheno_Data_ALL_201807 Query.xlsx')
ind3 <- match(stnumber,dat3$ST_Number)
pdat <- cbind(pdat,
control_subtype=dat2$Major.category[ind2],
control_subtype2=dat3$`Sub-type_Control`[ind3])
#-------------------------------------------#
pdat$control_subtype <- as.character(pdat$control_subtype)
pdat$control_subtype[pdat$control_subtype2=="Healthy volunteers from the general population"] <- "Healthy volunteer"
pdat$control_subtype <- factor(pdat$control_subtype, levels=
c("Benign adnexal cyst",
"Benign breast changes",
"Benign endometrial changes",
"Endometriosis",
"Ovarian Fibroma",
"Ovarian Teratoma",
"Pelvic floor dysfunction",
"Unspecified gynaecological problems",
"Uterine Fibroids",
"Healthy volunteer"))
ind <- !is.na(pdat$control_subtype)
n.factor <- length(levels(pdat$control_subtype[ind]))
pdat <- pdat[ind,]
head(pdat)
dat <- pdat %>%
mutate(`WID-BC-index` = index,
subtype = control_subtype) %>%
select(`WID-BC-index`, subtype)
rownames(dat) <- c()
head(dat)
table(dat$subtype)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-5b.txt")
#----------------------------------
# Suppl Fig 6
#----------------------------------
# a -------------------------------
theme_set(
theme_minimal() +
theme(legend.position = "top")
)
load("~/Dropbox/index-dev/archive/biobank-QC/4-calc-intensity/1-output/cin_intensity_cont.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/4-calc-intensity/2-output/ki_intensity_cont.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/4-calc-intensity/3-output/pros_intensity_cont.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/4-calc-intensity/5-output/this_intensity_cont.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/3-storage/3-output/cin_storage.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/3-storage/4-output/pros_storage.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/3-storage/1-output/ki_storage.Rdata")
load("~/Dropbox/index-dev/archive/biobank-QC/3-storage/2-output/thrc_storage.Rdata")
thrc_storage$col_to_receive[730] <- NA
#------ cin: storge & intensity
ind <- match(rownames(cin_intensity_cont), cin_storage$basename)
cin_storage <- cin_storage[ind,]
#identical(rownames(cin_intensity_cont), cin_storage$basename)
cin <- cbind(cin_storage, cin_intensity_cont)
#------ pros: storge & intensity
ind <- match(rownames(pros_intensity_cont), pros_storage$basename)
pros_storage <- pros_storage[ind,]
#identical(rownames(pros_intensity_cont), pros_storage$basename)
pros <- cbind(pros_storage, pros_intensity_cont)
#------ ki: storge & intensity
ind <- match(rownames(ki_intensity_cont), ki_storage$basename)
ki_storage <- ki_storage[ind,]
#sum(is.na(ind))
#head(rownames(ki_intensity_cont))
#head(ki_storage$basename)
ki <- cbind(ki_storage[,1:5], ki_intensity_cont)
ki <- ki[!is.na(ki$basename),]
#------ 3C: storge & intensity
ind <- match(rownames(this_intensity_cont), thrc_storage$basename)
thrc_storage <- thrc_storage[ind,]
#sum(is.na(ind))
#head(rownames(this_intensity_cont))
#head(thrc_storage$basename)
thrc <- cbind(thrc_storage[,1:4], this_intensity_cont)
thrc <- thrc[!is.na(thrc$basename),]
total <- as.numeric(nrow(cin) + nrow(pros) + nrow(ki) + nrow(thrc))
df <- data.frame(intensities = c(cin$oob_red, pros$oob_red, ki$oob_red, thrc$oob_red,
cin$oob_grn, pros$oob_grn, ki$oob_grn, thrc$oob_grn,
cin$typeI_red, pros$typeI_red, ki$typeI_red, thrc$typeI_red,
cin$typeI_grn, pros$typeI_grn, ki$typeI_grn, thrc$typeI_grn,
cin$typeII_red, pros$typeII_red, ki$typeII_red, thrc$typeII_red,
cin$typeII_grn, pros$typeII_grn, ki$typeII_grn, thrc$typeII_grn),
type = c(rep("oob_red", total), rep("oob_grn", total), rep("typeI_red", total),
rep("typeI_grn", total), rep("typeII_red", total), rep("typeII_grn", total)),
dataset = c(rep("CIN_dis", nrow(cin)), rep("CIN_val", nrow(pros)), rep("KI_val", nrow(ki)), rep("3C_dis", nrow(thrc)),
rep("CIN_dis", nrow(cin)), rep("CIN_val", nrow(pros)), rep("KI_val", nrow(ki)), rep("3C_dis", nrow(thrc)),
rep("CIN_dis", nrow(cin)), rep("CIN_val", nrow(pros)), rep("KI_val", nrow(ki)), rep("3C_dis", nrow(thrc)),
rep("CIN_dis", nrow(cin)), rep("CIN_val", nrow(pros)), rep("KI_val", nrow(ki)), rep("3C_dis", nrow(thrc)),
rep("CIN_dis", nrow(cin)), rep("CIN_val", nrow(pros)), rep("KI_val", nrow(ki)), rep("3C_dis", nrow(thrc)),
rep("CIN_dis", nrow(cin)), rep("CIN_val", nrow(pros)), rep("KI_val", nrow(ki)), rep("3C_dis", nrow(thrc)))
)
fill <- scale_fill_manual(labels = c("oob_red", "oob_grn", "typeI_red", "typeI_grn", "typeII_red", "typeII_grn"),
values=c("darkorange", "yellowgreen", "tomato1", "palegreen3", "firebrick3", "seagreen4"))
df$type <- factor(df$type, levels = c("oob_red", "oob_grn", "typeI_red", "typeI_grn", "typeII_red", "typeII_grn"))
df$dataset <- factor(df$dataset, levels = c("CIN_dis", "CIN_val", "KI_val", "3C_dis"))
head(df)
dat <- df %>%
filter(dataset %in% c("KI_val", "3C_dis")) %>%
mutate(set = case_when(dataset == "KI_val" ~ "Biobanked samples",
dataset == "3C_dis" ~ "Discovery samples"))
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-6a.txt")
# b -------------------------------
p <- c(0.99, seq(0.95,0.05,by = -0.05), 0.01)
er <- 500
eg <- 200
pdat <- data.frame(matrix(NA, nrow=200, ncol=length(p)))
for(i in 1:ncol(pdat)){
# true intensity
Us <-  p[i] * seq(0, 20000, length.out=200)
Ms <- (1-p[i]) * seq(0, 20000, length.out=200)
# observed intensity
U <- Us + er
M <- Ms + eg
# observed beta
pdat[,i] <- M/(M+U)
}
signal_to_noise2 <- (Us+Ms)/(er + eg)
pdatx <- pdat
head(pdatx)
pdatx <- cbind(signal_to_noise2, pdatx)
head(pdatx)
write.table(pdatx, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-6b.txt")
# c -------------------------------
datII <- readRDS("~/Dropbox/index-dev/archive/biobank-QC/5-beta-distributions/7-output/thrcII.Rds")
indx <- match(c('202081120239_R03C01','201496850045_R05C01'),colnames(datII))
head(datII)
high_snr <- data.frame(`Beta distribution` = datII[,indx[2]],
`Density` = datII[,1])
high_snr$snr <- "high SNR"
low_snr <- data.frame(`Beta distribution` = datII[,indx[1]],
`Density` = datII[,1])
low_snr$snr <- "low SNR"
dat <- rbind(high_snr, low_snr)
write.table(pdatx, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-6c.txt")
# d -------------------------------
pheno <- readRDS("~/Dropbox/index-dev/3C/0-discovery-preprocess/6-files/3C_pheno.Rds")
load("~/Dropbox/index-dev/3C/0-discovery-preprocess/8-files/rho_thrc.Rdata")
ind <- match(rownames(pheno), rownames(rho_thrc))
rho_thrc <- rho_thrc[ind,]
ind <- pheno$type %in% c('Control','Breast')
pdat <- rho_thrc[ind,]
beta_e <- 0.3309135
beta_s <- c(0.019570, 0.97065)
x <- seq(0,12,by=0.01)
y0 <- (x*beta_s[1] + beta_e)/(1+x)
y1 <- (x*beta_s[2] + beta_e)/(1+x)
df1 <- data.frame(x=x,y0=y0,y1=y1)
rownames(pdat) <- c()
write.table(pdat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-6d.txt")
# e -------------------------------
load("~/Dropbox/index-dev/KI-3C-validation/2-analysis/1-output/pdat.Rdata")
load("~/Dropbox/index-dev/KI-3C-validation/1-data-load/7-files/rho_ki.Rdata")
pdat <- pdat[pdat$cancer_type=='BC',]
ind <- match(rownames(pdat), rownames(rho_ki))
pdat <- rho_ki[ind,]
head(pdat)
rownames(pdat) <- c()
write.table(pdat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-6e.txt")
#----------------------------------
# Suppl Fig 7
#----------------------------------
load("~/Dropbox/index-dev/3C/1-BC/21-calibration/1-output/pdat_ext.Rdata")
dat <- pdat_ext %>%
mutate(`WID-BC-index (corrected prob)` = p_corrected,
type = type_binary) %>%
dplyr::select(`WID-BC-index (corrected prob)`, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-figure-7.txt")
#----------------------------------
# Suppl Table 1 (Epidemiological - discovery)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name == "3C_DISCOVERY" & type %in% c("Control", "Breast")) %>%
droplevels() %>%
mutate(type = case_when(type == "Control" ~ "Control",
type == "Breast" ~ "Cancer")) %>%
mutate(Age = age,
`Age at menarche` = age_at_menarche,
`Parity` = pregnancies_total,
`Age at first child live birth` = Pregnancy_age_1st_live_born,
`Menopausal status` = menopause,
`Age at menopause` = ifelse((menopause == "Post")==TRUE, last_period_age, NA),
`Oral contraceptive use` = OCP_current,
`Hormone replacement therapy` = HRT_current,
`Smoking` = smoking,
`First degree relatives with breast cancer` = first_degree_relatives_with_BC) %>%
dplyr::select(type, Age, `Age at menarche`, `Parity`, `Age at first child live birth`, `Menopausal status`, `Age at menopause`, `Oral contraceptive use`, `Hormone replacement therapy`, `Smoking`, `First degree relatives with breast cancer`)
rownames(dat) <- c()
# preM
pre <- dat[dat$`Menopausal status`=="Pre",]
table(pre$type, pre$`Oral contraceptive use`, useNA = "always")
# preM
post <- dat[dat$`Menopausal status`=="Post",]
table(post$type, post$`Hormone replacement therapy`, useNA = "always")
head(dat)
table(dat$type, dat$`First degree relatives with breast cancer`)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table1.txt")
#----------------------------------
# Suppl Table 2 (Epidemiological - ext)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name == "3C_VALIDATION_DATA" & dataset_name == "3C_external_validation" & type %in% c("Control", "Breast")) %>%
droplevels() %>%
mutate(type = case_when(type == "Control" ~ "Control",
type == "Breast" ~ "Cancer")) %>%
mutate(Age = age,
`Age at menarche` = age_at_menarche,
`Parity` = pregnancies_total,
`Age at first child live birth` = Pregnancy_age_1st_live_born,
`Menopausal status` = menopause,
`Age at menopause` = ifelse((menopause == "Post")==TRUE, last_period_age, NA),
`Oral contraceptive use` = OCP_ever,
`Hormone replacement therapy` = HRT_ever,
`Smoking` = smoking,
`First degree relatives with breast cancer` = first_degree_relatives_with_BC) %>%
dplyr::select(type, Age, `Age at menarche`, `Parity`, `Age at first child live birth`, `Menopausal status`, `Age at menopause`, `Oral contraceptive use`, `Hormone replacement therapy`, `Smoking`, `First degree relatives with breast cancer`)
rownames(dat) <- c()
# preM
pre <- dat[dat$`Menopausal status`=="Pre",]
table(pre$type, pre$`Oral contraceptive use`, useNA = "always")
# preM
post <- dat[dat$`Menopausal status`=="Post",]
table(post$type, post$`Hormone replacement therapy`, useNA = "always")
head(dat)
table(dat$type, dat$`First degree relatives with breast cancer`)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table2.txt")
#----------------------------------
# Suppl Table 3 (Clinical - 3C)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name %in% c("3C_DISCOVERY", "3C_VALIDATION_DATA") & type== "Breast") %>%
droplevels() %>%
mutate(set = case_when(experiment_name == "3C_DISCOVERY" ~ "Discovery set",
experiment_name == "3C_VALIDATION_DATA" ~ "External validation")) %>%
mutate(Age = age,
`Menopausal status` = menopause,
`T stage` = stage_T,
`N stage` = stage_N,
`Estrogen receptor status` = ER,
`Progesterone receptor status` = PR,
`HER2 status` = HER2,
`Grade` = grade,
`Histology` = histology) %>%
dplyr::select(set, Age, `Menopausal status`, `T stage`, `N stage`, `Estrogen receptor status`, `Progesterone receptor status`, `HER2 status`, `Grade`, `Histology`)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table3.txt")
#----------------------------------
# Suppl Table 4 (Epidemiological - OC)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name %in% c("3C_DISCOVERY", "3C_VALIDATION_DATA") & ((dataset_name == "3C_internal_validation" & type == "Control") | dataset_name %in% c("3C_internal_validation", "3C_training") & type == "Ovarian")) %>%
droplevels() %>%
mutate(type = case_when(type == "Control" ~ "Control",
type == "Ovarian" ~ "Cancer")) %>%
mutate(Age = age,
`Age at menarche` = age_at_menarche,
`Parity` = pregnancies_total,
`Age at first child live birth` = Pregnancy_age_1st_live_born,
`Menopausal status` = menopause,
`Age at menopause` = ifelse((menopause == "Post")==TRUE, last_period_age, NA),
`Oral contraceptive use` = OCP_current,
`Hormone replacement therapy` = HRT_current,
`Smoking` = smoking,
`First degree relatives with ovarian cancer` = first_degree_relatives_with_OC) %>%
dplyr::select(type, Age, `Age at menarche`, `Parity`, `Age at first child live birth`, `Menopausal status`, `Age at menopause`, `Oral contraceptive use`, `Hormone replacement therapy`, `Smoking`, `First degree relatives with ovarian cancer`)
rownames(dat) <- c()
# preM
pre <- dat[dat$`Menopausal status`=="Pre",]
table(pre$type, pre$`Oral contraceptive use`, useNA = "always")
# preM
post <- dat[dat$`Menopausal status`=="Post",]
table(post$type, post$`Hormone replacement therapy`, useNA = "always")
head(dat)
table(dat$type, dat$`First degree relatives with ovarian cancer`)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table4.txt")
#----------------------------------
# Suppl Table 5 (Clinical - OC)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name %in% c("3C_DISCOVERY", "3C_VALIDATION_DATA") & dataset_name %in% c("3C_internal_validation", "3C_training") & type == "Ovarian") %>%
droplevels() %>%
mutate(Age = age,
`Menopausal status` = menopause,
`T stage` = case_when(cancer_stage %in% c("IA", "Ib", "IB", "IC") ~ "T1",
cancer_stage %in% c("II", "IIA", "IIB", "IIC") ~ "T2",
cancer_stage %in% c("III", "IIIA", "IIIB", "IIIC", "IIIC-1", "IIIC-2") ~ "T3",
cancer_stage %in% c("IV", "IVA", "IVB") ~ "T4"),
`Grade` = grade,
`Histology` = histology) %>%
dplyr::select(Age, `Menopausal status`, `T stage`, `Grade`, `Histology`)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table5.txt")
#----------------------------------
# Suppl Table 6 (Epidemiological - EC)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name %in% c("3C_DISCOVERY", "3C_VALIDATION_DATA") & ((dataset_name == "3C_internal_validation" & type == "Control") | dataset_name %in% c("3C_internal_validation", "3C_training") & type == "Endometrial")) %>%
droplevels() %>%
mutate(type = case_when(type == "Control" ~ "Control",
type == "Endometrial" ~ "Cancer")) %>%
mutate(Age = age,
`Age at menarche` = age_at_menarche,
`Parity` = pregnancies_total,
`Age at first child live birth` = Pregnancy_age_1st_live_born,
`Menopausal status` = menopause,
`Age at menopause` = ifelse((menopause == "Post")==TRUE, last_period_age, NA),
`Oral contraceptive use` = OCP_current,
`Hormone replacement therapy` = HRT_current,
`Smoking` = smoking,
`First degree relatives with endometrial cancer` = first_degree_relatives_with_EC) %>%
dplyr::select(type, Age, `Age at menarche`, `Parity`, `Age at first child live birth`, `Menopausal status`, `Age at menopause`, `Oral contraceptive use`, `Hormone replacement therapy`, `Smoking`, `First degree relatives with endometrial cancer`)
rownames(dat) <- c()
# preM
pre <- dat[dat$`Menopausal status`=="Pre",]
table(pre$type, pre$`Oral contraceptive use`, useNA = "always")
# preM
post <- dat[dat$`Menopausal status`=="Post",]
table(post$type, post$`Hormone replacement therapy`, useNA = "always")
table(dat$Smoking, dat$type)
table(dat$type, dat$`First degree relatives with endometrial cancer`)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table6.txt")
#----------------------------------
# Suppl Table 7 (Clinical - EC)
#----------------------------------
load("~/Dropbox/index-dev/master-dataset/1-output/mdat.Rdata")
dat <- mdat %>%
filter(experiment_name %in% c("3C_DISCOVERY", "3C_VALIDATION_DATA") & dataset_name %in% c("3C_internal_validation", "3C_training") & type == "Endometrial") %>%
droplevels() %>%
mutate(Age = age,
`Menopausal status` = menopause,
`Stage` = cancer_stage,
`Grade` = grade,
`Histology` = histology) %>%
dplyr::select(Age, `Menopausal status`, `Stage`, `Grade`, `Histology`)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "suppl-table7.txt")
