Dear all, I have performed rma() on several CEL files. Here is the summary of this expression data : Min. 1st Qu. Median Mean 3rd Qu. Max. 2.674 5.021 6.150 6.144 7.151 14.420 The mean and variance of all arrays/columns are very close. A boxplot shows that the arrays are on comparable scales. However, these values are all positive. Do we take the log2 transformation followed by column median centering or use these rma output ? I am worried that this might make a difference if we use t-test or any other parametric tests. Can anyone help me ? Many thanks. Regards, Adai

they are already logged. rma doesnt subtract mm and uses a conservative background correction procedure so your minimum resulting expression values are usually around 8. thus the minimum log2 is around 3. if i were to do a t-test i would use the log scale expression values that rma gives you. On Wed, 20 Aug 2003, Adaikalavan Ramasamy wrote: > Dear all, > > I have performed rma() on several CEL files. Here is the summary of this > expression data : > > Min. 1st Qu. Median Mean 3rd Qu. Max. > 2.674 5.021 6.150 6.144 7.151 14.420 > > The mean and variance of all arrays/columns are very close. A boxplot > shows that the arrays are on comparable scales. > > However, these values are all positive. Do we take the log2 > transformation followed by column median centering or use these rma > output ? I am worried that this might make a difference if we use t-test > or any other parametric tests. > > Can anyone help me ? Many thanks. > > Regards, Adai > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor >