R/MEDIPS.CpGenrich.R

In HPCBio/MEDIPS-BioC: DNA IP-seq data analysis

##########################################################################
##Function calculates density of C's, G's, and CpGs in the given sequences as well as in the reference genome
##Returns CpG enrichment values for the given sequences.
##########################################################################
##Input:	bam file or tab (|) separated txt file "chr | start | stop  | strand"
##Param:	BSgenome, pattern (character), chromosomes, shift, extend
##Output:	CpGenrich results object
##Requires:	BSgenome, Biostrings, GenomicRanges, MEDIPS.Bam2GRanges, MEDIPS.txt2Granges, MEDIPS.GenomicCoordinates
##Modified:	11/10/2012
##Author:	Joern Dietrich, Lukas Chavez


MEDIPS.CpGenrich <-function(file=NULL, BSgenome=NULL, extend=0, shift=0, uniq=1e-3, chr.select=NULL, paired=F){

	## Proof correctness....
	if(is.null(BSgenome)){stop("Must specify a BSgenome library.")}
	
	## Read region file		
	fileName=unlist(strsplit(file, "/"))[length(unlist(strsplit(file, "/")))]
	path=paste(unlist(strsplit(file, "/"))[1:(length(unlist(strsplit(file, "/"))))-1], collapse="/") 
	if(path==""){path=getwd()}		
	if(!fileName%in%dir(path)){stop(paste("File", fileName, " not found in", path, sep =" "))}	

	dataset = get(ls(paste("package:", BSgenome, sep = "")))	

	if(!paired){GRange.Reads = getGRange(fileName, path, extend, shift, chr.select, dataset, uniq)}
	else{GRange.Reads = getPairedGRange(fileName, path, extend, shift, chr.select, dataset, uniq)}
	
	## Sort chromosomes
	if(length(unique(seqlevels(GRange.Reads)))>1){chromosomes=mixedsort(unique(seqlevels(GRange.Reads)))}
	if(length(unique(seqlevels(GRange.Reads)))==1){chromosomes=unique(seqlevels(GRange.Reads))}
	
	## Get chromosome lengths for all chromosomes within data set.
	cat(paste("Loading chromosome lengths for ",BSgenome, "...\n", sep=""))		

	chr_lengths=as.numeric(seqlengths(dataset)[chromosomes])

	ranges(GRange.Reads) <- restrict(ranges(GRange.Reads),+1)
	
	##Calculate CpG density for regions
	total=length(chromosomes)
	cat("Calculating CpG density for given regions...\n")  
	seq=matrix(unlist(IRanges::lapply(RangedData(GRange.Reads),function(x){		
		i=which(mixedsort(chromosomes)%in%names(x) )
		ranges(x)<-restrict(ranges(x),end=chr_lengths[which(chromosomes %in% names(x))])
		y=DNAStringSet(getSeq(dataset, names=space(x), start=start(x), end=end(x), as.character=TRUE))
		c(sum(as.numeric(vcountPattern("CG",y))),sum(as.numeric(vcountPattern("C",y))),sum(as.numeric(vcountPattern("G",y))),sum(as.numeric(width(y))),length(y))
		}
	),use.names=F),ncol=5,nrow=total,byrow=T)
  
	Value=colSums(seq)
	unused=length(GRange.Reads)-Value[5]
	if ( unused!=0 )cat(unused,"unused sequences, limits out of range\n")
	
	regions.CG=Value[1]
	regions.C=Value[2]
	regions.G=Value[3]
	all.genomic=Value[4]
	
	regions.relH=as.numeric(regions.CG)/as.numeric(all.genomic)*100
	regions.GoGe=(as.numeric(regions.CG)*as.numeric(all.genomic))/(as.numeric(regions.C)*as.numeric(regions.G))  
	
	##Calculate CpG density for reference genome
	cat(paste("Calculating CpG density for the reference genome", BSgenome, "...\n", sep = " "))	
	CG <- DNAStringSet("CG")
	pdict0 <- PDict(CG)
	params <- new("BSParams", X = dataset, FUN = countPDict, simplify = TRUE, exclude = c("rand", "chrUn"))
	genome.CG=sum(bsapply(params, pdict = pdict0))			
	params <- new("BSParams", X = dataset, FUN = alphabetFrequency, exclude = c("rand", "chrUn"), simplify=TRUE)
	alphabet=bsapply(params)
	genome.l=sum(as.numeric(alphabet))
 	
	genome.C=as.numeric(sum(alphabet[2,]))
	genome.G=as.numeric(sum(alphabet[3,]))
	genome.relH=genome.CG/genome.l*100
	genome.GoGe=(genome.CG*genome.l)/(genome.C*genome.G);
	
	enrichment.score.relH=regions.relH/genome.relH	
	enrichment.score.GoGe=regions.GoGe/genome.GoGe	
 	
	gc()
	return(list(genome=BSgenome, regions.CG=regions.CG, regions.C=regions.C, regions.G=regions.G, regions.relH=regions.relH, regions.GoGe=regions.GoGe, genome.C=genome.C, genome.G=genome.G, genome.CG=genome.CG, genome.relH=genome.relH, genome.GoGe=genome.GoGe, enrichment.score.relH=enrichment.score.relH, enrichment.score.GoGe=enrichment.score.GoGe))
}