I was surfing the blog and I stopped in this post C: Design and contrast question limma (additive or nested or duplicateCorrelation(). The guy had this experimental design:
Filename Sexe Diet Litter sample1 M LFD L37 sample2 M LFD L37 sample3 M LFD L49 sample4 M LFD L49 sample5 M LFD L50 sample6 M LFD L50 sample7 M WD L48 sample8 M WD L48 sample9 M WD L48 sample10 M WD L48 sample11 M WD L40 sample12 M WD L40 sample13 F LFD L49 sample14 F LFD L50 sample15 F LFD L37 sample16 F LFD L37 sample17 F LFD L37 sample18 F LFD L49 sample19 F WD L39 sample20 F WD L39 sample21 F WD L40 sample22 F WD L40 sample23 F WD L48 sample24 F WD L48
Aaron suggested to create a design matrix in this way:
design <- model.matrix(~0+Litter+paste(Sexe,Diet,sep="."))
Then he suggested to drop term number seven. I have doubt why dropping term number 7 the final two terms represent the average log-fold change for male mice over female, in the LFD or WD-fed mice. The design matrix would be:
Do the last two terms represent SDM.LFD-SDF.LFD and SDM.WD-SDF.LFD? Am I wrong?