Hello, I was wondering if anybody in this group can report experience with the combination of mas5 with Loess normalization for affy chips (in expresso: using mas for bgcorrect, pmcorrect and summary, and loess for the normalization). My data set contains two factors and 3 technical replicates. One factor seems to be very strong (lots of effect) whereas the other is rather weak. thanks for any advice, +regards, Arne