[BioC] Swirl Data Results

[Robert A Cribbie]

I am a newcomer to DNA microarrays and I have recently been playing around
with the swirl dataset that is available through the bioconductor R
package.

I have seen some results regarding the swirl data set on the internet that
have used one-sample empirical bayes moderated t-tests and the results
seem to come out as expected (i.e. the most differentially expressed
genes, bmp2 and dlx3, show up as the most differentially expressed).

I have also tried performing paired t-tests on the data (paired since they
share the same slide) after extracting the logR and logG values from the
normalized M and A values, but I get very different results. Very few
genes are differentially expressed with FDR control, and the top genes are
no longer the bmp2 and dlx3 genes. Has anyone else tried doing paired
t-tests on this data?

The commands I used were:

swirlnorm<-maNorm(swirl,norm="s")
M<-maM(swirlnorm)
A<-maA(swirlnorm)
logG<-(2*A-M)/2
logR<-(2*A+M)/2
newswirl<-cbind(logR[,1], logG[,1], logR[,2], logG[,2], logR[,3],
logG[,3], logR[,4], logG[,4])
classlabel<-c(1, 0, 0, 1, 1, 0, 0, 1)
tstat<-mt.teststat(newswirl,classlabel,test="pairt", nonpara='"n")
rawp<-2*(1-pt(abs(tstat),3))
result<-mt.rawp2adjp(rawp,proc=c("Bonferroni", "BH")
resultp<-mt.reject(result$adjp,seq(0,1,0.05))$r
resultp


Rob.