Hi, I am using limma to identify differentially expressed genes on 4 different platforms (CodeLink, Affymetrix, Agilent and an inhouse cDNA array), 2 RNA samples, 3 replicates for each on the one color microarrays, 6 replicates for the two color arrays. I used limma with fdr to adjust the p-value, and my result of differential genes from Affymetrix to the inhouse cDNA array was from 8000 to 500 differentially expressed probe sets. My question: is it reasonable to apply the same Thanks for some advice Anita Dr Anita Grigoriadis The Breakthrough Toby Robins Breast Cancer Research Centre, Institute of Cancer Research, 237 Fulham Road, London, SW3 6JB

> Date: Mon, 02 May 2005 23:11:57 +0100 > From: "Anita Grigoriadis" <anita.grigoriadis@icr.ac.uk> > Subject: [BioC] limma > To: <bioconductor@stat.math.ethz.ch> > > Hi, I am using limma to identify differentially expressed genes on 4 > different platforms (CodeLink, Affymetrix, Agilent and an inhouse cDNA > array), 2 RNA samples, 3 replicates for each on the one color > microarrays, 6 replicates for the two color arrays. I used limma with > fdr to adjust the p-value, and my result of differential genes from > Affymetrix to the inhouse cDNA array was from 8000 to 500 differentially > expressed probe sets. The number of genes coming up will depend on the precision of the platform, the number of replicates, the design of the experiment and the number of probes, so it is not surprising that the number of genes above a significance threshold will vary from one platform to another. > My question: is it reasonable to apply the same The same significance threshold? I guess it is if you're trying to identify differentially expressed probes. If your real purpose is to compare the platforms, this isn't the way I'd go about it. Gordon > Thanks for some advice > Anita > > > Dr Anita Grigoriadis > The Breakthrough Toby Robins > Breast Cancer Research Centre, > Institute of Cancer Research, > 237 Fulham Road, > London, SW3 6JB