Thank you. Yhe I would like to use RMA but for now I'm trying to repeat someone former results. The problem is that I am getting twice as many genes. And the expression values are extremely different. Why is the scaling so random and different from the one done by affymetrix MAS5? Cheers, Liat. Date: Thu, 12 Jul 2012 11:49:13 -0400 Subject: Re: [BioC] Affy mas5 scaling From: chenyao.bioinfor@gmail.com To: liats80@hotmail.com CC: bioconductor@r-project.org Hi Liat, Personally, I will use RMA instead of MAS5 to normalization affy data. It's quite normal when you use different parameter in normalization, you get different number of genes. But it's not a big problem. I would expect highly overlap of these two sets of genes. Jack 2012/7/12 Liat S <liats80@hotmail.com> Hi all, I have some raw data as well as Affymetrix MAS5 normalized data for oligonucleotide arrays. For the raw data I use the mas5 function of the affy package. >From then on I run the same code on both (removing AFFY controls, applying log, selecting genes over a certain fold change etc.). I end up with different numbers of genes... I have done A LOT of debugging (including doing the same actions manually in excel :o/ ) and am quite convinced now that the difference has to do with the normalization itself. The values of normalized data are very different, with the R normalized data having much higher values. I noticed that the mas5 function in the Affy package uses 500 as a scaling value. Can that be the source of the difference I see? Is there anyway to better approximate the MAS5 scaling? Thanks! Liat. [[alternative HTML version deleted]] _______________________________________________ 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]]