The example is a simple two group design where one group is treatment and the other is control. There should be no need to re-organize your files, but it important to provide a sheet / targets file giving the location of the files as well as relevant phenodata. You do that easiest by editing the 450k sheet so it reflects the file organization you are using, for example by using Excel. Regarding the reference, the user manual clearly (I hope) says that Illumina's normalization picks a single array as reference, and we have not been able to figure out how that is chosen. It should not matter, but the numerical results will differ a bit depending on the choice of reference. Whether or not that is an appropriate normalization procedure is something you will have to investigate, but I note that a number of papers have been published with alternatives, one of which (SWAN) is available in minfi. Please direct further questions to the bioc email list, which is cc'ed here. Best, Kasper D Hansen On Fri, Apr 5, 2013 at 10:50 AM, BH Tang <justice_woo@hotmail.com> wrote: > Hi Dr Hansen, > > I'm currently using your MINFI package in processing 450K data. Our > experiment samples (treatment/control) were put into different folder > groups, with the folder name group1, group2, etc. Each folder group > contains the both control and treatment sample data. > > While following your MINFI user guide, it only mentioned Group A and Group > B as a study case. Do the Groups A and B refer to the control and treatment > samples, respectively? > > For my case, do I need to group all those samples into the two folders, > and organize those samples' name and ID into the 450K sheet? > > In your user guide (Page 9), you adopt "reference=2" for normalization, > see below, > MSet.norm <- preprocessIllumina(RGset, bg.correct = TRUE, normalize = > "controls", reference = 2) > > For my case, which option should I choose to normalize the samples? > > Thanks a lot for your suggestion. > > Sincerely, > Ben > [[alternative HTML version deleted]]