Dear Gordon, I have a question about creating the proper design matrix for my data. I am using Affymetrix GeneChips and the study design is as follow: _ 40 old male and female exercised during 6 months three times a week. During the 6 months some individuals received protein supplement and some received placebo. The design is unbalanced (the protein placebo numbers are not equal , also male and female are not the same number). At both times( before exercise and after exercise) the Transcriptomics data for all individuals are available. Subjects time gender supplement subject-code 1 0 1 ? A 1 1 1 1 A 2 0 0 ? H 2 1 0 0 H 3 0 1 ? J 3 1 1 0 J .. .. 40 0 0 ? K 40 1 0 1 K In the above design time indicates before (0) and after(1) exercise, gender(0=male, 1=female), supplement(0=placebo,1=protein),subject_code is the indication of the same individual. ? :I wonder that for supplement how I should indicate the protein , placebo at baseline (time=0) because at time=0 ,nobody got any supplement. Primary questions to be addressed: 1) Is there any effect of exercise( time) in old people regarding there gene expression profile? (Which genes are affected based on exercise?) 2) Is there any effect of protein supplementation ? 3) Any interaction between protein supplement and exercise effect? 4) Sex effect ? So to do so, I need to run paired analysis in limma, but till now the design does not work properly. May I ask you to give me some help how to design as I should consider several factors( sex, time(before and after exercise), supplement, within individual variation,...interactions) Thank you so much ion advance Regards Pari [[alternative HTML version deleted]]