Hi I'm using limma for an affymetrix dataset. For summarize which is the most relavant differential expressed genes I use classifyTestsF() and I extract up-regulated and down-regulated genes. At the same time i look at the results of the fit with topTable() (sorted by B). But the genes are different. Is it possible? Why? Is B (probability log odds) a god estimator? What is the better solution, to summarize my results, between B and F-statistics? Thanks Daniela Marconi