[BioC] understanding GOstats p-value&In-Reply-To=E759ADD5-DCF1-4F5C-B6D6-475031AF81CE at fhcrc.org

[Chunyan Liu]

Dear Jim,

I have a similar question, but it's for limma. I noticed when I want to make contrasts among the three levels (A, B, and C) of a factor, limma does not make adjustment for the multiple contrasts and only make adjustment within each contrast by whatever method I specify in the decideTests function. Should I adjust for the multiple contrast? Does your argument about the dependency among the GO terms apply to this question? The following are the R code for the contrasts. Thanks.

contrast.matrix <- makeContrasts(A-C,B-C,0.5*(A+B)-C, levels=design)
fit2 <- contrasts.fit(fit, contrast.matrix)
fit2 <- eBayes(fit2)
results <- decideTests(fit2,p.value=0.05,adjust.method="fdr")

Chunyan 

Biostatistician I
Center for Epidemiology and Biostatistics
Cincinnati Children's Hospital Medical Center
513-636-9763