Dear all, I have several experimental batches that I would like to compare in one huge analysis. Preferably, I would like to use limma for the analysis. I am not worried that there will be a batch-specific bias: in fact, one of the goal of the analysis will be to find out to what extent the results from on batch are confirmed by an independent analysis for another cohort. Also, the general idea of the study is to evaluate the practical feasibility of such cross-cohort approaches. However, the problem is that some of the experiments were done on two-color arrays. I could split the channels manually, but that is clearly not optimal. I know there is the possibility of analysing channels separately using intraspotCorrelation(): corfit <- intraspotCorrelation( M, design ) fit <- lmscFit( M, design, correlation= corfit$consensus.correlation ) Is there a way of using this approach to correct and separate the two channels? Kind regards, j. -- -------- Dr. January Weiner 3 --------------------------------------