rownames(dat) <- c()
colnames(dat) <- c("immune cell proportion", "TGCA-BC-index", "type", "tissue")
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-1h.txt")
rm(list=ls);gc()
#---------------------------------
# Figure 3
#---------------------------------
rm(list=ls())
col <- pal_npg(palette = c("nrc"), alpha = 1)(7)
# source plot functions
dir_list <- dir('~/Dropbox/src/manuscript-plots')
for(d in dir_list){
source(paste('~/Dropbox/src/manuscript-plots/',d,sep=""))
}
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/1-output/pdat.Rdata")
n <- 29000
# BRCA specific
source('~/Dropbox/src/plot-tools/plot_correlation.R')
source('~/Dropbox/index-dev/3C/1-BC/7-analysis/13-output/plot_box.R')
# a   ---------------------------
ind <- pdat$dataset=='External_validation'
type_new <- rep('Control',nrow(pdat))
type_new[pdat$type=='Breast'] <- 'Breast cancer'
pdat$type <- factor(type_new, levels=c('Control','Breast cancer'))
dat <- droplevels(pdat[ind,])
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index) %>%
mutate(`immune cell proportion` = ic) %>%
select(`WID-BC-index`, `immune cell proportion`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-3a.txt")
# b   ---------------------------
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-3b.txt")
# c   ---------------------------
load("~/Dropbox/index-dev/KI-3C-validation/2-analysis/1-output/pdat.Rdata")
pdat_ki <- pdat[pdat$cancer_type=='BC',]
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$type %in% c('Control','Breast')
pdat2 <- data.frame(stn = as.numeric(c(rho_thrc$rho[ind], pdat_ki$rho)),
dataset = c(rep('3C',sum(ind)), rep('KI',nrow(pdat_ki))))
pdat2 <- pdat2[!is.na(pdat2[,1]),]
colnames(pdat2) <- c("signal-to-noise", "dataset")
write.table(pdat2, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-3c.txt")
# d   ---------------------------
load("~/Dropbox/index-dev/KI-3C-validation/2-analysis/1-output/pdat.Rdata")
ind <- (pdat$cancer_type == 'BC') &
(!(pdat$stage %in% c('T0', 'T1', 'Tis')) |
!(pdat$nodal %in% c('N0'))) &
!(pdat$person_sqnr %in% c(9, 17, 23, 73, 148, 164, 244, 347, 367, 376, 419, 438, 443, 551))
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_BC) %>%
mutate(`signal-to-noise` = rho) %>%
select(`WID-BC-index`, `signal-to-noise`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-3d.txt")
# e  ---------------------------
ind <-
(pdat$cancer_type == 'BC') &
!(pdat$stage %in% c('T2', 'T3', 'T4b')) &
!(pdat$nodal %in% c('N1', 'N1', 'N2', 'N3', 'Nx')) &
!(pdat$person_sqnr %in% c(9, 17, 23, 73, 148, 164, 244, 347, 367, 376, 419, 438, 443, 551))
gdat <- pdat[ind,]
rownames(gdat) <- c()
gdat <- gdat %>%
mutate(`WID-BC-index` = index_BC) %>%
mutate(`signal-to-noise` = rho) %>%
select(`WID-BC-index`, `signal-to-noise`, type)
write.table(gdat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-3e.txt")
#---------------------------------
# Figure 4
#---------------------------------
rm(list=ls())
col <- pal_npg(palette = c("nrc"), alpha = 1)(7)
# source plot functions
dir_list <- dir('~/Dropbox/src/manuscript-plots')
for(d in dir_list){
source(paste('~/Dropbox/src/manuscript-plots/',d,sep=""))
}
# a  ---------------------------
# cervical internal validation set
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/12-output/pdat_BC.Rdata")
# entire buccal 3c set
load("~/Dropbox/index-dev/3C/1-BC/10-buccal/1-output/pdat.Rdata")
type_new <- rep('Control',nrow(pdat))
type_new[pdat$type=='Breast'] <- 'Breast cancer'
pdat$type <- factor(type_new, levels=c('Control','Breast cancer'))
# load buccal phenotype
dat <- read.csv('~/Dropbox/index-dev/3C/0-buccal-preprocess/1-files/BC_RISK_BUCCALS_PHENO_240119.csv')
basename <- dat$buccal_basename
beta_names <- rownames(pdat)
ind <- match(beta_names,basename)
cbasename <- dat$cervical_basename[ind]
ind1 <- match(cbasename, rownames(pdat_BC))
# save subsets of each pdat dataset
pdat_buccal <- pdat[!is.na(ind1),]
pdat_cervix <- pdat_BC[ind[!is.na(ind1)],]
dat <- pdat_buccal %>%
mutate(`WID-BC-index` = index) %>%
mutate(`immune cell proportion` = ic) %>%
select(`WID-BC-index`, `immune cell proportion`, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-4a.txt")
# b  ---------------------------
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-4b.txt")
# c  ---------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/16-output/pdat.Rdata")
ind <- pdat$dataset=='Training' & pdat$type=='Control'
type_new <- rep('Control',nrow(pdat))
type_new[pdat$type=='Ovarian'] <- 'Ovarian cancer'
pdat$type <- factor(type_new, levels=c('Control','Ovarian cancer'))
pdat1 <- pdat[!ind,]
dat <- pdat1 %>%
mutate(`WID-BC-index` = index) %>%
mutate(`immune cell proportion` = ic) %>%
select(`WID-BC-index`, `immune cell proportion`, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-4c.txt")
# d  ---------------------------
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-4d.txt")
# e  ---------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/17-output/pdat.Rdata")
ind <- pdat$dataset=='Training' & pdat$type=='Control'
type_new <- rep('Control',nrow(pdat))
type_new[pdat$type=='Endometrial'] <- 'Endometrial cancer'
pdat$type <- factor(type_new, levels=c('Control','Endometrial cancer'))
pdat1 <- pdat[!ind,]
dat <- pdat1 %>%
mutate(`WID-BC-index` = index) %>%
mutate(`immune cell proportion` = ic) %>%
select(`WID-BC-index`, `immune cell proportion`, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-4e.txt")
# f  ---------------------------
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-4f.txt")
#---------------------------------
# Figure 5
#---------------------------------
rm(list=ls())
col <- pal_npg(alpha = 1)(7)
# source plot functions
dir_list <- dir('~/Dropbox/src/manuscript-plots')
for(d in dir_list){
source(paste('~/Dropbox/src/manuscript-plots/',d,sep=""))
}
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/1-output/pdat.Rdata")
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation') &
(pdat$type=='Control')
TITLE <- 'Internal + External Validation Controls'
RANGE <- c(-1.7,2.4)
# a  ---------------------------
source('~/Dropbox/src/plot-tools/plot_age.R')
index_new <- pdat$index
index_new[pdat$dataset=='External_validation'] <- pdat$index_tr_version[pdat$dataset=='External_validation']
pdat$index <- index_new
type_new <- rep('Control',nrow(pdat))
type_new[pdat$type=='Breast'] <- 'Breast cancer'
pdat$type <- factor(type_new, levels=c('Control','Breast cancer'))
dat <- droplevels(pdat[pdat$dataset!='Training',])
dat <- dat %>%
mutate(`WID-BC-index` = index) %>%
select(`WID-BC-index`, age, type)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5a.txt")
# b  ---------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/1-output/pdat.Rdata")
source('~/Dropbox/src/plot-tools/consensus_roc.R')
snp <- read.csv(file='~/Dropbox/index-dev/3C/1-BC/16-SNP/from-olivia-13mar2019/Combined_summary_stats_beta_BC_4C_303SNPs.csv')
ind <- pdat$dataset=='Internal_validation'
pdat_val <- pdat[ind,]
basename <- rownames(pdat_val)
ind <- match(basename, as.character(snp$Cervical_Cell_Basename))
prs_val <- snp$X303_SCORE[ind]
centre <- snp$u_ucldepartment[ind]
ind0 <- !is.na(prs_val)
pdat1 <- data.frame(index=pdat_val$index[ind0],
prs=prs_val[ind0],
type=pdat_val$type[ind0])
pdat1$type <- factor(pdat1$type, levels=c("Breast","Control"))
pdat1$type <- plyr::revalue(pdat1$type, c("Breast"="Breast cancer"))
pdat1$type <- factor(pdat1$type, levels=c('Control','Breast cancer'))
colnames(pdat1) <- c("WID-BC-index", "PRS", "type")
write.table(pdat1, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5b.txt")
# c  ---------------------------
dat <- data.frame(index = snp$X303_SCORE,
type = snp$Type,
age=snp$Age_at_consent)
colnames(dat) <- c("PRS", "type", "age")
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5c.txt")
# d  ---------------------------
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/19-output/mdat.Rdata")
source("~/Dropbox/src/manuscript-plots/plot_two_rocs.R")
dat <- mdat %>%
filter(type %in% c("Control", "Breast")) %>%
mutate(type = factor(type, levels = c("Control", "Breast"))) %>%
mutate(`WID-BC-index` = index_BC_discovery_version) %>%
mutate(`WID-BC-index w/o mQTLs` = index_BC_nomqtl) %>%
select(type, `WID-BC-index`, `WID-BC-index w/o mQTLs`) %>%
droplevels()
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5d.txt")
# e  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation') &
(pdat$type=='Control')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
select(`WID-BC-index`, BMI, `type`)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5e.txt")
# f  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation') &
(pdat$type=='Control')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
mutate(`1st degree relatives with BC` = first_degree_relatives_with_BC) %>%
select(`WID-BC-index`, `1st degree relatives with BC`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5f.txt")
# g  ---------------------------
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
mutate(`Age at menarche` = age_at_menarche) %>%
select(`WID-BC-index`, `Age at menarche`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5g.txt")
# h  ---------------------------
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
mutate(`Pregnancy age at 1st live born` = Pregnancy_age_1st_live_born) %>%
select(`WID-BC-index`, `Pregnancy age at 1st live born`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5h.txt")
# i  ---------------------------
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
mutate(`Menopausal status` = menopausal) %>%
select(`WID-BC-index`, `Menopausal status`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5i.txt")
# j  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation') &
(pdat$type=='Control') &
(pdat$menopausal=='Post')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
select(`WID-BC-index`, HRT, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5j.txt")
# k  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
filter(stage_T != "") %>%
mutate(`T stage` = stage_T) %>%
select(`WID-BC-index`, `T stage`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5k.txt")
# l  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
filter(stage_N != "") %>%
mutate(`N stage` = stage_N) %>%
select(`WID-BC-index`, `N stage`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5l.txt")
# m  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
filter(grade != "") %>%
mutate(`Grade` = grade) %>%
select(`WID-BC-index`, `Grade`, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5m.txt")
# n  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
select(`WID-BC-index`, ER, PR, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5n.txt")
# o  ---------------------------
ind <- (pdat$dataset=='Internal_validation' | pdat$dataset=='External_validation')
dat <- pdat[ind,]
rownames(dat) <- c()
dat <- dat %>%
mutate(`WID-BC-index` = index_tr_version) %>%
select(`WID-BC-index`, HER2, type)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-5o.txt")
#---------------------------------
# Figure 6
#---------------------------------
# a  ---------------------------
load("~/Dropbox/index-dev/ENCODE/2-BC/2-output/sdat.Rdata")
dat <- sdat[sdat$tissue!="tissue",]
dat <- dat %>%
mutate(`WID-BC-index` = index) %>%
select(`WID-BC-index`, type, nm)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-6a.txt")
# b  ---------------------------
load("~/Dropbox/index-dev/ENCODE/2-BC/1-output/pdat.Rdata")
type_tmp <- rep("Epithelial",nrow(pdat))
#type_tmp[pdat$type=='M'] <- "Mesenchymal"
type_tmp[pdat$type=='non_epithelial'] <- "Non-epithelial"
pdat$type <- factor(type_tmp, levels=c('Non-epithelial','Epithelial'))
pdat <- pdat[pdat$tissue=='tissue',]
pdat$type <- factor(pdat$type, levels=c('Epithelial','Non-epithelial'))
dat <- pdat %>%
mutate(`WID-BC-index` = index) %>%
select(`WID-BC-index`, type, Fat)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-6b.txt")
# c  ---------------------------
load("~/Dropbox/index-dev/3C/1-BC/18-TFBS-analysis/16-output/delta-betas.Rdata")
hypo <- names(delta_3c[delta_3c < -0.02])
hypo_3c <- delta_3c[names(delta_3c) %in% hypo]
pdat2 <- data.frame(x = delta_3c,
y = delta_atrisk)
colnames(pdat2) <- c("delta beta in cervical samples", "delta beta in breast tissue at risk")
write.table(pdat2, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-6c.txt")
# d  ---------------------------
load("~/Dropbox/index-dev/3C/1-BC/18-TFBS-analysis/16-output/brca_risk.Rdata")
dat <- pdat %>%
mutate(`Mean PRBS methylation` = mean,
type = brca.flag) %>%
select(`Mean PRBS methylation`, type)
head(dat)
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "figure-6d.txt")
#---------------------------------
# Table 1
#---------------------------------
# BC internal/external set
source('~/Dropbox/src/manuscript-plots/table_or.R')
library(epitools)
library(MASS)
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/1-output/pdat.Rdata")
bc <- pdat %>%
filter(dataset != "Training") %>%
mutate(set = case_when(dataset == "Internal_validation" ~ "Breast cancer internal validation",
dataset == "External_validation" ~ "Breast cancer external validation")) %>%
mutate(`WID-BC-index` = index) %>%
dplyr::select(`WID-BC-index`, type, set)
rm(pdat)
# OC
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/16-output/pdat.Rdata")
ind <- pdat$dataset=='Training' & pdat$type=='Control'
type_new <- rep('Control',nrow(pdat))
type_new[pdat$type=='Ovarian'] <- 'Ovarian cancer'
pdat$type <- factor(type_new, levels=c('Ovarian cancer','Control'))
pdat <- pdat[!ind,]
oc <- pdat %>%
mutate(set = "Ovarian cancer") %>%
mutate(`WID-BC-index` = index) %>%
dplyr::select(`WID-BC-index`, type, set)
# EC
load("~/Dropbox/index-dev/3C/1-BC/7-analysis/17-output/pdat.Rdata")
ind <- pdat$dataset=='Training' & pdat$type=='Control'
pdat <- pdat[!ind,]
dat <- read.csv('~/Dropbox/index-dev/3C/0-discovery-preprocess/1-files/PostM_WID_EPIC_LIST_1680_Masterlist_with_phenotypic_data_20180806.csv')
ind <- match(rownames(pdat),dat$Basename)
age <- dat$Age_at_consent[ind]
menopause <- dat$Menopausal_Status[ind]
pdat <- cbind(pdat, age, menopause)
pdat$type <- factor(pdat$type, levels=c('Endometrial','Control'))
ec <- pdat %>%
mutate(set = "Endometrial cancer") %>%
mutate(`WID-BC-index` = index) %>%
dplyr::select(`WID-BC-index`, type, set)
# combine + save
dat <- rbind(bc, oc, ec)
rownames(dat) <- c()
write.table(dat, sep = "\t",
quote = FALSE,
row.names = FALSE,
file = "table1.txt")
library(impute)
version(impute)
impute
help(package = "impute")
help(package = "minfi")
help(package = "ChAMP")
help(package = "glmnet")
help(package = "fgsea")
help(package = "EpiDISH")
help(package = "pROC")
help(package = "iterativeBMA")
BiocManager::install("iterativeBMA")
help(package="iterativeBMA")
help(package="rms")
help("TCGAbiolinks")
library(TCGAbiolinks)
library(package="TCGAbiolinks")
help(package="GEOquery")
help(package="ChAMP")
help(package="minfi")
help(package="impute")
help(package="EpiDISH")
help(package="TCGAbiolinks")
help(package="GEOquery")
help(package="pROC")
# source plot functions
dir_list <- dir('~/Dropbox/src/manuscript-plots')
for(d in dir_list){
source(paste('~/Dropbox/src/manuscript-plots/',d,sep=""))
}
View(plot_box)
