Ah, yes, re-reading the original question, this is the correct answer. Filtering genes by biological annotation (e.g. just TFs) should be done after running limma. On Tue Jun 10 19:47:59 2014, Paul Geeleher wrote: > Limma borrows information across all genes to better estimate variance > for each gene, so your results should in theory be more accurate if > you run the analysis for all genes on your arrays and export the > p-values for the set of genes in which you are interested at the end > of the analysis. > > Paul > > On Tue, Jun 10, 2014 at 3:34 AM, Bas van Gestel > <shc.van.gestel at="" gmail.com=""> wrote: >> Dear all, >> >> Currently, I'm comparing the gene expression of three cell types to >> identify differentially expressed genes. For this purpose, the limma >> package seems ideal. However, for one of the analyses, I would like to >> focus only on the transcription factors of the cell types. This implies >> that I only consider about 990 genes. Is this enough to perform an analysis >> with limma, or does it require a larger number of genes at minimum to give >> reliable results? >> >> Kind regards, >> Bas van Gestel >> >> [[alternative HTML version deleted]] >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor at r-project.org >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor > > >