I am analysing Affy chip data for differential expression. Previously I have done this using the 'limma' package without weighting the data. More recently I have been assessing quality control using the weights from the affyPLM command fitPLM, which I have found to show up spatial variation on Affy chips very clearly. I would like to use these weights in the subsequent analysis. Does it make any sort of sense to feed the affyPLM weights into the limma lm.series analysis? Alternatively, is there some way to use the limma eBayes technique on the fitPLM output? -- Tom Price: telephone +1 215 898 0489