Can I suggest that limma package returns the posterior probabilities along with the log odds ratio, B ? I am currently using the following codes to get the posterior values (unless I am reading the documents wrong). fit <- lmFit(onebatch, design) fit2 <- contrasts.fit( fit, contrast.mat ) fit2 <- eBayes(fit2) results <- topTable(fit2) results$post.prob <- exp(results$B) / ( 1 + exp(results$B) ) Hopefully this will get people to start thinking in terms of probabilities and less on p-values. Thank you. Regards, Adai