Find most significant mutations for survival

maftools.R

require(maftools)

maf <- GDCquery_Maf("COAD",pipelines = "mutect2")
clinic <- GDCquery_clinic("TCGA-COAD")
colnames(clinic)[1] <- "Tumor_Sample_Barcode"
clinic$Overall_Survival_Status <- 1
clinic$Overall_Survival_Status[which(clinic$vital_status != "dead")] <- 0 
clinic$time <- clinic$days_to_death
clinic$time[is.na(clinic$days_to_death)] <- clinic$days_to_last_follow_up[is.na(clinic$days_to_death)]
acc <- read.maf(maf = maf, clinicalData = clinic,isTCGA = T)

genes <- getGeneSummary(acc)$Hugo_Symbol[1:20]
tbl <- plyr::adply(genes,1, function(x){
  surv = mafSurvival(maf = acc, 
                     genes = x, 
                     fn = x,
                     time = 'time', 
                     Status = 'Overall_Survival_Status', 
                     isTCGA = TRUE)
  pvalue <- surv$labels$subtitle
  pvalue
},.progress = "text")
tbl$X1 <- genes
colnames(tbl) <- c("gene","pvalue")
tbl$pvalue <- gsub("P-value: ","",tbl$pvalue )
tbl$pvalue <- as.numeric(tbl$pvalue)
tbl <- tbl[order(tbl$pvalue),]