Hi James, thanks for your help. The problem with this interaction is, that I find nearly none DE transcripts in our RNA-Seq experiment (3 replicates) while based on our microarray results we expect something in the range of a few hundreds... Am I correct that this interaction finds significance if a) the expression of the two groups is sig. different, or b) the fold-change ratio is different (ie. C1 goes from 100 to 1000 when treating with Fusarium compared to mock while C2 goes just from 50 to 100) or) C1 is upregulated while C2 is downregulated when treated with Fusarium and vice versa. My problem is to interpret the contrasts correctly, besides the basic contrasts. So basically whats the difference between this two contrasts: Diff_C1C2=(C1F-C1M)-(C2F-C2M) and Diff_C1C2_new = (C1F+C2F)/2-(C1M+C2M)/2 ? Will the second contrast just find significance if both groups are either up or downregulated? Thanks once more for your help! All the best Christian >>> "James W. MacDonald" <jmacdon@uw.edu> 7/18/2012 2:55 >>> Hi Christian, On 7/17/2012 4:32 AM, Christian Ametz wrote: > Dear members, > > I'm struggling to create the contrast matrix for our experiment comprising of 4 genotypes in treated and untreated conditions: > > The genotypes are labelled C1 to C4 > The conditions are treated (F) and untreated (M) > I set up the contrast matrix like this: > > cont.matrix.30<- makeContrasts( > C1_FvsM=C1F-C1M, > C2_FvsM=C2F-C2M, > C3_FvsM=C3F-C3M, > C4_FvsM=C4F-C4M, > Diff_C1C2=(C1F-C1M)-(C2F-C2M), > levels=design) > > > The first four "normal" contrasts estimating the response of treated/untreated genotypes are working great. My problem is the "double contrast" Diff_C1C2 where I want to find those genes that show a difference in response (treated vs untreated) between the two genotypes C1 and C2. The technical term for that contrast is an interaction, and you have set it up correctly. So what is the problem? Best, Jim > > > I would gladly accept any help! > > Thanks > Christian > > > > > > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -- James W. MacDonald, M.S. Biostatistician University of Washington Environmental and Occupational Health Sciences 4225 Roosevelt Way NE, # 100 Seattle WA 98105-6099 [[alternative HTML version deleted]]