Dear All,

I am using following code for my expression analysis from the edgeR package.

data = read.table("mirna.readcout", header=T, row.names=1, com='')

col_ordering = c(1,2,3,4,5,6)
rnaseqMatrix = data[,col_ordering]
rnaseqMatrix = round(rnaseqMatrix)
rnaseqMatrix = rnaseqMatrix[rowSums(rnaseqMatrix)>=2,]
conditions = factor(c(rep("Set1", 3), rep("Set2", 3)))

exp_study = DGEList(counts=rnaseqMatrix, group=conditions)
exp_study = calcNormFactors(exp_study)
exp_study = estimateCommonDisp(exp_study)
exp_study = estimateTagwiseDisp(exp_study)

et = exactTest(exp_study)

tTags = topTags(et,n=NULL)
write.table(tTags, file='mirna.Set1_vs_Set2.edgeR.DE_results', sep=' ', quote=F, row.names=T)

I would like to get the normalised value Per Group (ie. Set1 & Set2) which is used in the exactTest. How can I get that? In otherwords the data which is used to calculate logFC?.

Looking forward for your reply, Thank you very much for your time.

Best Regards,

Manoharan

The simplest way would be to do something like this:

design <- model.matrix(~0 + conditions)
fit <- glmFit(exp_study, design)

... then the values of fit$coefficients/log(2) represent the normalized average log-expression in each group. Subtracting one from the other should recover the log-fold change (this will not be sorted, though, so you'll have to run topTags with sort.by='none' for them to match up).

Dear Aaron Lun,

Thank you very much.

Best Regards,

Manoharan