Hello Michael, I am sure more experienced members will have better answers, but here is my $0.02. Your doubt is similar to the doubt I had before. So, hopefully I can answer your question. You are right about the relation between AveExpr and Amean. However, AveExpr is not same as value of the fitted coefficients. AveExpr is the average log intensity for a particular probe across microarrays. The fitted coefficients will be the same as logFC value from the topTable output. The coefficients are named as logFC in topTable output as the change in log intensities when treatment changes from 1 to 2 is equal to the value of coefficient (which is same as logFC). Please correct me if I misunderstood your question. Thanks and Best Regards, S. On Tue, Nov 30, 2010 at 2:40 PM, Michael Muratet <mmuratet@hudsonalpha.org>wrote: > Dear Colleagues > > I have a question about how the AveExpr value gets calculated in limma > topTable. Actually, it appears to get calculated in lmFit, e.g., > > from lmFit: > $Amean > uc007aet.1 uc007aeu.1 uc007aev.1 uc007aew.1 uc007aex.1 > 0.8269211 1.1462191 -2.3369824 -6.0639390 -8.4696942 > > from topTable: > acomps.txt:uc007aet.1 -1.07454088439694 0.826921104052945 -0.83217169279942 > 0.413619779606768 0.999993313025155 -5.09985857376248 > bcomps.txt:uc007aet.1 -0.590082310688743 0.826921104052945 > -0.559690866321207 0.580942531774888 0.999999437256152 -4.75598562188099 > > So AveExpr = Amean > > But I have looked at the 'average log-intensity' for the raw values and for > the fitted coefficients for log10, log2 and log and looked at mean(log(x)) > and log(mean(x)) and I can't get 0.8269211. I can post the data if need be. > > Moreover, many of the fitted coefficients are less than zero and > log(mean(x)) is defined but not mean(log(x)). > > Can anyone tell me how Amean (and therefore AveExpr) gets calculated? > > Thanks > > Mike > > > > Michael Muratet, Ph.D. > Senior Scientist > HudsonAlpha Institute for Biotechnology > mmuratet@hudsonalpha.org > (256) 327-0473 (p) > (256) 327-0966 (f) > > Room 4005 > 601 Genome Way > Huntsville, Alabama 35806 > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > [[alternative HTML version deleted]]