Normalise with median fold change

June 01, 2012, 05:55:55 PM

Recently there was a paper published with suggested that median fold change was a suitable normalization method for urine and other samples that are effected by dilutions.

Code: [Select]

normalize.medFC <- function(mat) {
	# Perform median fold change normalisation
	#           X - data set [Variables & Samples]
	medSam <- apply(mat, 1, median)
	medSam[which(medSam==0)] <- 0.0001
	mat <- apply(mat, 2, function(mat, medSam){
		medFDiSmpl <- mat/medSam
		vec<-mat/median(medFDiSmpl)
		return(vec)
	}, medSam)
	return (mat)
}

On the faahKO dataset it would work something like the following:

Code: [Select]

cdfpath <- system.file("cdf", package = "faahKO")
cdffiles <- list.files(cdfpath, recursive = TRUE,full=T)
faahko <- xcmsSet(cdffiles)
faahko <- group(faahko)
faahko <- retcor(faahko)
faahko <- fillPeaks(faahko)

report <- cbind(groups(faahko), normalize.medFC(groupval(faahko, "medret", "into")) )
head(report)
write.csv(report, file="reportFile.csv")

(1) Veselkov, K. A.; Vingara, L. K.; Masson, P.; Robinette, S. L.; Want, E.; Li, J. V.; Barton, R. H.; Boursier-Neyret, C.; Walther, B.; Ebbels, T. M. D.; Pelczer, I.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Anal. Chem. 2011, 83, 5864–5872.