Hi, I am just doing a straightforward analysis of mouse affy data using affy,gcrma and limma It is a simple 2 group design with not much complication to it. So this is a minor problem that I have been scratching my head since the beggining of the week, and cannot figure out the answer, so hopefully someone can help me. After I read my data, normalize and analyze as follows: library (affy) library (limma) library (gcrma) Data <- ReadAffy() eset <- gcrma(Data) treatment<-c("SD","SD","NSD","NSD","NSD","NSD","NSD","NSD","NSD","NSD" ,"NSD","SD","SD","SD","SD","SD","SD") design <-model.matrix(~factor(treatment)) colnames(design) <- c("SD","NSD") fit <- lmFit(eset, design) fit <- eBayes(fit) I just wanted to get all the normalized data for the GEO report, however I get strange results regarding the scale of the expression values if I use "exprs" from Biobase. Log Expression values range from ~1-4 if I do: esetNL= exprs (eset) esetlog=log(esetNL,2) write.table(esetlog, file="Myfile_gcrma_matrix.txt", quote=F, sep="\t") The dynamic range of log expression values I have usually observed ranges from ~3-14, and it is what I get if I do: >write.table(topTable(fit, coef=2, adjust="fdr", sort.by="A",number=50000),file="limma_SD.txt", sep="\t") although that only gives me the average expression values. Is the data extracted by "exprs" already in log2 scale? can someone tell why I am seeing such difference in scale/dynamic range? Am I not using "exprs" correctly? thanks very much in advance Lucia [[alternative HTML version deleted]]

On 06/08/10 00:32, Lucia Peixoto wrote: > (...) > although that only gives me the average expression values. > Is the data extracted by "exprs" already in log2 scale? can someone tell why > I am seeing such difference in scale/dynamic range? Am I not using "exprs" > correctly? > > The data coming out of rma/gcrma is already log2-transformed. L.