Hello everyone I was wondering if anyone has used a tool to adjust for batch effects. I am comparing two experiments that were done the same way over two years. eg. Population 1 Treatment Control Year 1 Population 1 Treatment Control Year 2. The same experiment, just repeated in different years. I normalise all the cel files together (from both years) using GCRMA and look for DE genes with limma, (with arrayWeights and later eBayes adjustment before multiple testing with BH FDR). I incorporate both years data into the linear model and have Year as a factor as well. I am interested in probesets shared between both years (differentially expressed in treatment compared to control) but I see differences as well which are interesting. I would suspect some differences could simply be attributed to batch effects when processing the chips rather than an actual sample influence. I'm am not sure how much an effect normalising all the cel files together removes this year-by-year processing influence. Does anyone recommend of any tools that could correct for these effects? Is it neccessary? I have come across ComBat ( http://statistics.byu.edu/johnson/ComBat/Abstract.html). To be honest, I would not know which one would be the best, and any advice/experiences will be appreciated. Thank you. -- Regards, Sherosha [[alternative HTML version deleted]]