Dear Sam, No, there isn't a way to model a factorial experiment that is more powerful than the methods indicated in the edgeR User's Guide. The methods in the User's Guide already fully account for the fact that the double mutant has expression dependent on each mutation and any interaction. Best wishes Gordon > Date: Wed, 25 Dec 2013 20:01:14 -0600 > From: Sam McInturf <smcinturf at="" gmail.com=""> > To: "bioconductor at r-project.org" <bioconductor at="" r-project.org=""> > Subject: [BioC] edgeR complex partial multifactorial design > > Hello everyone, > I need to make a design matrix, and I can use the approach outlined in > 3.3.1 of the edgeR users guide, but I think that there may be a better way > to conduct the analysis. I have 36 samples in a 3x2x2 incomplete factorial > design. I have three genotypes (wild type, a single mutant, and a double > mutant), two tissues (roots and shoots), and two treatments (no trt, trt). > the mutants are m1 and m1/m2. One approach would be to group the treatment > groups into a single variable Group, as is done in the edgeR guide, and > then follow the example, extracting interaction terms to find the > differences between each genotypes response. > > There should be a different way to model this problem that is more / as > powerful as the previous method (right?). Because in the double mutant the > gene expression is going to have a component that is dependent on each > mutation and then the interaction of each mutation. > > So I am interested in trying to estimate the genes that are responsive to > treatment and dependent on the introduction of mut2 into the mut1 > background (for tissues independently) > > Does such a method exist, how can I do it / what should I read to > understand how to do it? > > Thanks, > > -- > Sam McInturf ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}