arg<-c()
#variables input
n<-length(arg)
if((n %% 3)!=0){
	stop()
}else{
	filelist<-arg[1:(n/3)]
	namelist<-arg[(n/3+1):(n*2/3)]
	survlist<-arg[(n*2/3+1):n]
}
library("randomForestSRC")
library("survival")
binary_trans<-function(x){
	for(i in 3:ncol(x)){
		x[,i]<-(as.numeric(x[,i])-mean(as.numeric(x[,i])))/sd(as.numeric(x[,i]))
	}
	return(x)
}
cox.analysis<-function(genes,coeff,dataset,name,type){
	dataset$sampleID<-rownames(dataset)
	dataset<-dataset[,c(ncol(dataset),1:(ncol(dataset)-1))]
	genes<-c(genes, "OS", "OS_IND")
	genes<-as.data.frame(genes,stringsAsFactors = FALSE)
	sample<-dataset$sampleID
	workplace<- as.data.frame(sample, stringsAsFactors=F)
	RFName <- names(dataset)
	for(i in 1:(length(genes[,1])-2)) {
	  g <- genes[i,1]
	  index <- which(RFName == g)
	  workplace <- cbind(dataset[[index]], workplace)
	  names(workplace)[1] <- g
	}
	workplace$OS<-dataset$OS
	workplace$OS_IND<-dataset$OS_IND
	workplace$sample<-NULL
	OS<-as.numeric(workplace$OS)
	OS_IND<-as.numeric(workplace$OS_IND)
	workplace$OS<-NULL
	workplace$OS_IND<-NULL
	score<-as.data.frame(sample, stringsAsFactors=FALSE)
	for(i in 1:length(workplace[1,])) {
	  score1 <- workplace[,i]*(as.numeric(coeff[i,1]))
	  score <- cbind(score1, score)
	}
	score$sample<-NULL
	riskscore<-apply(score, 1, sum)
	rr<-summary(as.numeric(riskscore))
	median <- as.numeric(rr[3])
	medianB <- as.numeric(as.numeric(riskscore)>=median)
	ss <- survdiff(Surv(OS, OS_IND) ~ medianB)
	p_os <- 1-pchisq(ss$chisq, 1)
	sf <- survfit(Surv(OS, OS_IND) ~ medianB)
	pdf(file=paste(name,type,"survial.pdf",sep="_"))
	plot(sf,xlab="survival time",ylab="survival probabilty",col=1:2,axes=F,lwd=2)
	axis(1);axis(2)
	legend("bottomleft",paste("p=",p_os),col=2)
	dev.off()
	data.write<-cbind(sample,workplace,OS, OS_IND, riskscore, medianB, stringsAsFactors=FALSE)
	data.write<-as.data.frame(data.write, stringsAsFactors=FALSE)
	write.csv(data.write, file = paste("data",name,type,"_exp_profile.csv",sep="_"), row.names = F, quote = F)
}
for(i in 1:length(filelist)){
	x<-read.table(filelist[i],header=T,sep="\t")
	rownames(x)<-x$sampleID
	x<-x[,-1]
	if(i==1){
			total.out <- var.select(Surv(OS, OS_IND) ~ ., x, ntree=1000,importance = TRUE,
					method = "vh",  nrep = 100, nstep = 100)
		sink("rfsrc_summary_total.out_rfsrc_glance.txt")
		print(total.out$rfsrc.refit.obj)
		sink()
		write.table(total.out$rfsrc.refit.obj$err.rate, file="rfsrc_total.out_every_erro_rate.txt", quote=F, sep="\t")
		pdf("rfsrc_var_selected_plot_A.pdf", width=5.9, height=6.0)
		plot(total.out$err.rate)
		dev.off()
		importance<-total.out$varselect
		vimportance<-as.data.frame(importance, stringsAsFactors=FALSE)
		write.table(vimportance, file="rfsrc_total.out_topvars_importance.txt", sep="\t", quote=F, row.names=T)
		Gene<-total.out$topvars
		indexes<-c()
		for(j in 1:length(Gene)){
			index<-which(Gene[j]==colnames(x))
			indexes<-c(indexes,index)
		}
		workplace<-x[,indexes]
		workplace$OS<-x$OS
		workplace$OS_IND<-x$OS_IND
		coxm<-coxph(Surv(OS, OS_IND) ~ ., workplace)
		b<-as.data.frame(summary(coxm)$coefficient, stringsAsFactors=FALSE)
		write.table(b, file="coefficient_of_select_gene.txt", sep='\t', quote=F, row.names=F)
		x<-binary_trans(x)
		x$sampleID<-rownames(x)
		x<-x[,c(ncol(x),1:(ncol(x)-1))]
		cox.analysis(Gene,b,x,namelist[i],survlist[i])
	}else{
		x<-binary_trans(x)
		cox.analysis(Gene,b,x,namelist[i],survlist[i])
	}
}