Dear Carlos, There is a section in the limma User's Guide called "Time Course Experiments". Although the example given there is for Affymetrix arrays, have you tried adapting it to your situation? Best wishes Gordon >Date: Wed, 5 Jul 2006 01:42:27 +0000 (UTC) >From: Carlos Calderon <ccaldero at="" ira.cinvestav.mx=""> >Subject: [BioC] analysis of timeseries with treatment effect >To: bioconductor at stat.math.ethz.ch > >Bioconductor comunity: >I have been trying to find the best way to answer the classical questions for >a timeseries with two treatment effect. Altough I have been reading I am not >able to find an example about the best matrix design and contrasts >in order to >answer the above questions. >If I use modelMatrix() using the targets frame I have got a matrix >design that >makes me able to identify genes with differential expression between >treatments in a single time point. But about the second question, how the >treatment changes across the timeseries? I'm not sure how to treat >it. I guess >that I need to analyze the interaction between time and treatment but my so >poor experience using linear models and making contrasts limits my analysis, >so I think I am not fully taking advantage of all the data. >The experiment is a two-color time series experiment, 4 time points, 2 >treatments, 16 arrays: > > t1--->t2--->t3--->t4--->t1 > II II II II >c4<---c1<---c2<---c3<---c4 > >2hybs per time point for direct comparisons (getting 8 slides) plus the other >8 slides to make hybs across times, as indicated in the representation. >Thank you in advance.