Hello all, I'm working with Affy GeneChips, and would like to know if anyone has suggestions regarding the following: I will only be doing downstream analysis on a subset of the probes, for example: ~12,000 out of ~45,000 that I can actually map to a transcript with a 3'UTR. - Would it make sense to normalize (or a 2nd normalization?) using only this subset of probes? - Would it be better to use limma with the subset or the total set? I imagine the model would be more robust with all probes, but the p-value correction would then be more stringent. Any pointers to other reading material would be greatly appreciated. Thanks! Cei The Wellcome Trust Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE.