R/normalise.R

In mQTL.NMR: Metabolomic Quantitative Trait Locus Mapping for 1H NMR data

normalise <-
function(X,method,refIdx=1,noiseInt=c(11,12))
{
# Removing dilutions between biofluid samples (normalisation of spectra)

# Normalisation and scaling of spectra profiles

# X [observation dimension]
# method - Constant sum (method<-'CS')
#        - Constant noise (method<-'CN')
#        - Qoutient probabilistic method (method<-'PQN')
#        - Linear baseline normalisation (method<-'LBN')
#        - Auto-scaling (method<-'AS')
#        - Pareto scaling (method<-'PS')
# L. Hedjazi, ICAN, 2014

if (nargs()<2)
{
    stop('incorrect number of input parameters')
}
if (!is.matrix(X)){X<-as.matrix(X)}

obs<-dim(X)[1]
dimm<-dim(X)[2]


# Normalisation

factors<-rep(NaN,obs)


switch(method,
       CS={
       print("Start constant sum normalisation:")
       factors<-apply(X,1,sum)
       med<-median(factors)
       data.cs<-med*(X/factors)
       return(list(Sp=data.cs, factors=factors))},

       CN={
       print("Start constant noise normalisation:")
       ppm<-as.numeric(sub("e\\.","e-",sub("ppm_","",sub("ppm_\\.","-", colnames(X)))))
       st<-which(ppm>=noiseInt[1] & ppm<=noiseInt[2])
       start<-min(st)
       end<-max(st)
       factors<-apply(X[,start:end],1,sd)
       data.cn<-(X/factors)
       return(list(Sp=data.cn, factors=factors))},

       LBN={
       print("Start linear baseline normalisation:")
       linear.baseline <- apply(X,2,median)
       baseline.mean <- mean(linear.baseline)
       sample.means <- apply(X,1,mean)
       factors <- baseline.mean/sample.means
       data.lbn<-(X*factors)
       return(list(Sp=data.lbn, factors=factors))},
       PQN={
       print("Start probabilistic quotient normalisation:")
       if (missing(refIdx)) {
       reference <- apply(X,2,median)
       }else{reference<-X[refIdx,]}

       X[0==X]<-0.00000001
       quotient <- t(X)/reference
       quotient.median <- apply(quotient,2,median)
       data.pqn<-(X/quotient.median)
       return(list(Sp=data.pqn, factors=quotient.median, NormSp=reference))},
       AS={
       print("Start Auto-scaling:")
       centered.data <- t(X) - apply(X,2,mean)
       scaling.auto <- apply(X,2,sd)
       auto.data <- t(centered.data/scaling.auto)
       return(list(Sp=auto.data, factors=scaling.auto))},
       PS={
       print("Start Pareto scaling:")
       centered.data <- t(X) - apply(X,2,mean)
       scaling.pareto <- sqrt(apply(X,2,sd))
       pareto.data <- t(centered.data/scaling.pareto)
       return(list(Sp=pareto.data, factors=scaling.pareto))})
}