Though I have spent many hours reading the limma user's guide, and pored over chapter 23 of the Gentleman, Carey et al Bioconductor book, I have many more questions than answers regarding limma, and its use with our data. I come from a programming background, with some math training (in the distant past); I am sure many crucial statistical issues elude me. I am looking for help. Our experiment seeks to discover Plasmodium falciparum up-regulated genes in peripheral blood samples drawn from four preganant women and eight children, all of whom have malaria. A schematic of the design may be seen at http://gaggle.systemsbiology.net/pshannon/limma.html In this diagram, each line represents two slides which are simple dye- swap pairs. Due to a shortage of mRNA, only some of the many possible comparisons were made. I have performed some separate limma analyses of each dye-swapped pair. Many of the same genes -- the top 10 or so -- appear in most slides. These genes make sense biologically. Now I wish to probe more deeply, analyzing the data in aggregate to discover more from the data. In the BioC book, sections 23.2 ('Data representations') and section 23.9 ('Direct two-color designs') suggest to me that we have a direct 2-color design. As a first step, I hope to find out if that is true. If so -- or if not -- I am sure the answer will lead to more questions... Thank you, - Paul Shannon Seattle Biomedical Research Institute & Institute for Systems Biology