R/nullp.R

In goseq: Gene Ontology analyser for RNA-seq and other length biased data

#############################################################################
#Description: Generates the weighting curve that is used to generate the length corrected NULL distribution.
#A spline is fit to obtain a functional relationship between gene length and likelihood of differential exrpession
#Notes: By default genome and id are used to fetch length data from GeneLenDataBase, but the length of each gene can be supplied with bias.data
#Author: Matthew Young
#Date Modified: 20/12/2010


nullp = function(DEgenes, genome, id, bias.data=NULL, plot.fit=TRUE){
	#Input Checking
	if(!is.null(bias.data) & length(bias.data)!=length(DEgenes)){
		stop("bias.data vector must have the same length as DEgenes vector!")
	}
	#Factors cause strange things to happen, remove them if they exist
	bias.data=unfactor(bias.data)
	DEgenes=unfactor(DEgenes)

	#Fetch length data from geneLenDataBase
	if(is.null(bias.data)){
		bias.data=getlength(names(DEgenes),genome,id)
	}

	#Fit a spline to the binary vector of DE calls vs gene length
	#May not have bias data for some of the entries, return NA at those positions
	pwf=rep(NA,length(DEgenes))
	w=!is.na(bias.data)
	pwf[w]=makespline(bias.data[w],DEgenes[w])

	#Make a data frame which contains all the data used to make the fit and the fit itself
	out=data.frame(DEgenes=DEgenes,bias.data=bias.data,pwf=pwf,stringsAsFactors=FALSE)
	rownames(out)=names(DEgenes)

	#Plot the PWF if the arument has been specified
	if(plot.fit){
		plotPWF(out)
	}
	out
}