Naomi: ok. so I got carried away trying to be a rabble rowser: I misread something you said..about avoiding FDR stuff. You were actually saying that you avoid the statistical selection schemes when making gene lists... on account of they didn't work so well (less correlated than just using means/fold changes). and i think this comment is ....a warning sign, too...not for you, but for all of us in the community. Estimating single gene variances is hard and frought with peril. The thing to do is to borrow strength from BIOLOGICALLY RELATED genes. cashaw