Mercurial > repos > sblanck > smagexp

#!/usr/bin/env Rscript
# setup R error handling to go to stderr
options( show.error.messages=F, error = function () { cat( geterrmessage(), file=stderr() ); q( "no", 1, F ) } )

# we need that to not crash galaxy with an UTF8 error on German LC settings.
loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")

library("optparse")

##### Read options
option_list=list(
		make_option("--input",type="character",default="NULL",help="list of rdata objects containing eset objects"),
		make_option("--result",type="character",default=NULL,help="text file containing result of the meta-analysis"),
		make_option("--htmloutput",type="character",default=NULL,help="Output html report"),
		make_option("--htmloutputpath",type="character",default="NULL",help="Path of output html report"),
		make_option("--htmltemplate",type="character",default=NULL,help="html template)")
);

opt_parser = OptionParser(option_list=option_list);
opt = parse_args(opt_parser);

if(is.null(opt$input)){
	print_help(opt_parser)
	stop("input required.", call.=FALSE)
}

#loading libraries

suppressPackageStartupMessages(require(metaMA))
suppressPackageStartupMessages(require(affy))
suppressPackageStartupMessages(require(annaffy))
suppressPackageStartupMessages(require(VennDiagram))
suppressPackageStartupMessages(require(GEOquery))

listInput <- trimws( unlist( strsplit(trimws(opt$input), ",") ) )

listfiles=vector()
listfilenames=vector()

for (i in 1:length(listInput))
{
	inputFileInfo <- unlist( strsplit( listInput[i], ';' ) )
	listfiles=c(listfiles,inputFileInfo[1])
	listfilenames=c(listfilenames,inputFileInfo[2])
}

cargs <- commandArgs()
cargs <- cargs[(which(cargs == "--args")+1):length(cargs)]
nbargs=length(cargs)
listfiles=vector()
listfilenames=vector()
for (i in seq(1,nbargs-6,2)) {
	listfiles=c(listfiles,cargs[[i]])
	listfilenames=c(listfilenames,cargs[[i+1]])
}
#mod<-cargs[[length(cargs) - 6]]
outputfile <- opt$result
result.html = opt$htmloutput
html.files.path=opt$htmloutputpath
result.template=opt$htmltemplate

alpha=0.05

#print(comparison)

listData=lapply(listfiles,read.table)
orderData=lapply(listData, function(x) x[order(x[1]), ])
rawpval=lapply(orderData,function(x) x[6])
rawpval=lapply(rawpval, function(x) as.numeric(unlist(x)))

DE=list()
DE=lapply(orderData, function(x) ifelse(x[7]<=0.05,1,0))

FC=list()
FC=lapply(orderData, function(x) x[3])

DE=as.data.frame(DE)
colnames(DE)=listfilenames
FC=as.data.frame(FC)
colnames(FC)=listfilenames
# the comparison must only have two values and the conds must
# be a vector from those values, at least one of each.

#if (length(comparison) != 2) {
#	stop("Comparison type must be a tuple: ", cargs[length(cargs) - 8])
#}

sink("/dev/null")
dir.create(html.files.path, recursive=TRUE)
#library(DESeq)
#library(HTSFilter)

#DE=list()
#FC=list()
#i=1

# Open the html output file
#file.conn = file(diag.html, open="w")

#writeLines( c("<html><body>"), file.conn)

# Perform deseq analysis on each study
#for(i in 1:length(listfiles))
#{
#  f=listfiles[i]
#  fname=listfilenames[i]
#  study_name=unlist(strsplit(fname,"[.]"))[1]
#  print(paste0("study.name ",study_name))
#  d <- read.table(f, sep=" ", header=TRUE, row.names=1)
#  conds<-sapply(strsplit(colnames(d),"[.]"),FUN=function(x) x[1])
#  if (length(unique(conds)) != 2) {
#  	warning(as.data.frame(strsplit(colnames(d),"[.]")))
#  	stop("You can only have two columns types: ", paste(conds,collapse=" "))
#  }
#  if (!identical(sort(comparison), sort(unique(conds)))) {
#  	stop("Column types must use the two names from Comparison type, and vice versa.  Must have at least one of each in the Column types.\nColumn types: ", cargs[2], "\n", "Comparison type: ", cargs[3])
#  }
#  if (length(d) != length(conds)) {
#  	stop("Number of total sample columns in counts file must correspond to the columns types field.  E.g. if column types is 'kidney,kidney,liver,liver' then number of sample columns in counts file must be 4 as well.")
#  }
#
#  cds <- newCountDataSet(d, conds)
#  cds <- estimateSizeFactors(cds)
#
#  cdsBlind <- estimateDispersions( cds, method="blind" )
#
#  if (length(conds) != 2) {
#  	cds <- estimateDispersions( cds )
#  	norep = FALSE
#  }
#
#  if (length(conds) == 2) {
#  	cds <- estimateDispersions( cds, method=method, sharingMode=mod, fitType="parametric" )
#  	norep = TRUE
#  }
#
#  filter<-HTSFilter(cds, plot=FALSE)
#  cds.filter<-filter$filteredData
#  on.index<-which(filter$on==1)
#
#  res<-as.data.frame(matrix(NA,nrow=nrow(cds),ncol=ncol(cds)))
#  nbT <- nbinomTest(cds.filter, comparison[1], comparison[2])
#  colnames(res)<-colnames(nbT)
#  res[on.index,]<-nbT
#  #write.table(res[order(res$padj), ], file=outputfile, quote=FALSE, row.names=FALSE, sep="\t")
#
#
#  temp.pval.plot = file.path( html.files.path, paste("PvalHist",i,".png",sep=""))
#  png( temp.pval.plot, width=500, height=500 )
#  hist(res$pval, breaks=100, col="skyblue", border="slateblue", main="")
#  dev.off()
#
#  writeLines( c("<h2>P-value histogram for ",study_name,"</h2>"), file.conn)
#  writeLines( c("<img src='PvalHist",i,".png'><br/><br/>"), file.conn)
#
#  #on enregistre la p-value
#  rawpval[[study_name]]<-res$pval
#  DE[[study_name]]<-ifelse(res$padj<=alpha,1,0)
#  FC[[study_name]]<-res$log2FoldChange
#
#  i=i+1
#}


# combinations
library(metaRNASeq)
fishcomb<-fishercomb(rawpval, BHth=alpha)
warning(length(rawpval))
invnormcomb<-invnorm(rawpval, nrep=c(8,8), BHth=alpha)
#DE[["fishercomb"]]<-ifelse(fishcomb$adjpval<=alpha,1,0)
#DE[["invnormcomb"]]<-ifelse(invnormcomb$adjpval<=alpha,1,0)

signsFC<-mapply(FC,FUN=function(x) sign(x))
sumsigns<-apply(signsFC,1,sum)
commonsgnFC<-ifelse(abs(sumsigns)==dim(signsFC)[2],sign(sumsigns),0)

DEresults <- data.frame(DE=DE,"DE.fishercomb"=ifelse(fishcomb$adjpval<=alpha,1,0),"DE.invnorm"=ifelse(invnormcomb$adjpval<=alpha,1,0))

unionDE <- unique(c(fishcomb$DEindices,invnormcomb$DEindices))
FC.selecDE <- data.frame(DEresults[unionDE,],FC[unionDE,],signFC=commonsgnFC[unionDE])
keepDE <- FC.selecDE[which(abs(FC.selecDE$signFC)==1),]

fishcomb_de <- rownames(keepDE)[which(keepDE[,"DE.fishercomb"]==1)]
invnorm_de <- rownames(keepDE)[which(keepDE[,"DE.invnorm"]==1)]
indstudy_de = list()
for (i in 1:length(listfiles)) {
	currentIndstudy_de = rownames(keepDE)[which(keepDE[,i]==1)]
	indstudy_de[[listfilenames[i]]]=currentIndstudy_de
}

IDDIRRfishcomb=IDD.IRR(fishcomb_de,indstudy_de)
IDDIRRinvnorm=IDD.IRR(invnorm_de,indstudy_de)

#conflits<-data.frame(ID=listData[[1]][rownames(DEresults),1],Fishercomb=DEresults[["DE.fishercomb"]],Invnormcomb=DEresults[["DE.invnorm"]],sign=commonsgnFC)
conflits<-data.frame(ID=listData[[1]][rownames(DEresults),1],DE=DEresults,FC=FC,signFC=commonsgnFC)
#write DE outputfile
write.table(conflits, outputfile,sep="\t",,row.names=FALSE)
library(VennDiagram)
DE_num=apply(DEresults, 2, FUN=function(x) which(x==1))
venn.plot<-venn.diagram(x=as.list(DE_num),filename=NULL, col="black", fill=1:length(DE_num)+1,alpha=0.6)
temp.venn.plot = file.path( html.files.path, paste("venn.png"))
png(temp.venn.plot,width=500,height=500)
grid.draw(venn.plot)
dev.off()

library(jsonlite)
matrixConflits=as.matrix(conflits)
datajson=toJSON(matrixConflits,pretty = TRUE)
summaryFishcombjson=toJSON(as.matrix(t(IDDIRRfishcomb)),pretty = TRUE)
summaryinvnormjson=toJSON(as.matrix(t(IDDIRRinvnorm)),pretty = TRUE)


#vennsplit=strsplit(result.venn,split="/")[[1]]
#venn=paste0("./",vennsplit[length(vennsplit)])


vennFilename="venn.png"
vennFile=file.path(html.files.path,vennFilename)
htmlfile=readChar(result.template, file.info(result.template)$size)
htmlfile=gsub(x=htmlfile,pattern = "###DATAJSON###",replacement = datajson, fixed = TRUE)
htmlfile=gsub(x=htmlfile,pattern = "###FISHSUMMARYJSON###",replacement = summaryFishcombjson, fixed = TRUE)
htmlfile=gsub(x=htmlfile,pattern = "###INVSUMMARYJSON###",replacement = summaryinvnormjson, fixed = TRUE)
htmlfile=gsub(x=htmlfile,pattern = "###VENN###",replacement = vennFilename, fixed = TRUE)
write(htmlfile,result.html)

#library(VennDiagram)
#flog.threshold(ERROR)
#
##venn.plot<-venn.diagram(x = c(res[c(1:(length(res)-3))],meta=list(res$Meta)),filename = v, col = "black", fill = c(1:(length(res)-2)), margin=0.05, alpha = 0.6,imagetype = "png")
#dir.create(result.path, showWarnings = TRUE, recursive = FALSE)
#
#showVenn<-function(liste,file)
#{
#	venn.plot<-venn.diagram(x = liste,
#			filename = vennFilename, col = "black",
#			fill = 1:length(liste)+1,
#			margin=0.05, alpha = 0.6,imagetype = "png")
##	png(file);
##	grid.draw(venn.plot);
##	dev.off();
#
#}
#
#l=list()
#for(i in 1:length(esets))
#{
#	l[[paste("study",i,sep="")]]<-res[[i]]
#}
#l[["Meta"]]=res[[length(res)-1]]
#showVenn(l,vennFile)
#file.copy(vennFilename,result.path)


#writeLines( c("<h2>Venn Plot</h2>"), file.conn)
#writeLines( c("<img src='venn.png'><br/><br/>"), file.conn)
#writeLines( c("</body></html>"), file.conn)
#close(file.conn)
#print("passe6")
#sink(NULL)
author	sblanck
date	Tue, 16 May 2017 01:53:32 -0400
parents	93451f832736
children	ef7d98f9eb51