Hi, I am using rma to normalize a set of Affymetrix exon 1.0 ST arrays. I have done this both with the oligo package and with the Affymetrix Power Tools to get equivalent results. I find that the first principal component is completely determined by one of the quality measures, namely pos_control_mean (the mean of the positive control probes as returned by Affymetrix Power Tools) to the extent that pos_control_mean and PCA 1 have an R^2 of 0.93. There is no sign of any outliers. It appears to me that the rma normalization is not removing the brightness effect at all. I find this surprising. Is this what other people are seeing? I am regressing the normalized data on the pos_control_mean and doing further analysis on the residuals. Bye R