一文解决如何提取TCGA配对表达矩阵

#######################################################
## 
#######################################################

rm(list=ls())



load( "mRNA_exprSet.Rda")


metadata <- data.frame(colnames(mRNA_exprSet ))

colnames(metadata) <- 'barcode'


for (i in 1:length(metadata[,1])) {
  num <- as.numeric(as.character(substring(metadata[i,'barcode'],14,15)))
  if (num == 1 ) {metadata[i,2] <- "T"}
  if (num != 1) {metadata[i,2] <- "N"}
}



names(metadata)[2] <- 'Barcode'

table(metadata$Barcode)

N <- subset(metadata, metadata$Barcode == 'N')

N$barcode <- substr(x=N$barcode, start = 1, stop = 12)


#######################################################
## 
#######################################################



dt <- as.data.frame(t(mRNA_exprSet))

dt[1:3,1:3]

dt$Barcode <- rownames(dt)

dt$sample  <- substr(x=dt$Barcode, start = 1, stop = 12)

dt <- subset(dt, select=c('Barcode', 'sample'))

dt <- dt[which(dt$sample %in% N$barcode),]

table(dt$sample)

df <- as.data.frame(table(dt$sample))

df <- subset(df, df$Freq > 1)

dt <- dt[which(dt$sample %in% df$Var1),]



mRNA_exprSet <- mRNA_exprSet[,which(colnames(mRNA_exprSet) %in% dt$Barcode)]



#######################################################
## 
#######################################################



metadata <- data.frame(colnames(mRNA_exprSet ))

colnames(metadata) <- 'barcode'


for (i in 1:length(metadata[,1])) {
  num <- as.numeric(as.character(substring(metadata[i,'barcode'],14,15)))
  if (num == 1 ) {metadata[i,2] <- "T"}
  if (num != 1) {metadata[i,2] <- "N"}
}

names(metadata)[2] <- 'Barcode'

table(metadata$Barcode)








select_colname <- function(expr,name){
  expr <- expr[,which(colnames(expr) %in% name[,1])]
  expr  <- expr[,order(names(expr))]   
  name <- name[which(name[,1] %in% colnames(expr)),]
  name <- name[order(name[,1]),]
  expr_name <- list( expr,name)
  return( expr_name)
}

datexpr <- select_colname(mRNA_exprSet ,metadata)


data <- datexpr[[1]]
metadata <- datexpr[[2]]

group <- metadata$Barcode

exprSet <- data

save(data, file = 'exprSet.Rdata')

save(group , file = 'group.Rdata')

##########################################################################################
## 
###########################################################################################