Hello, I'm trying to analyze a dataset in which there are repeated measures using edgeR. I have 48 RNA samples, collected from 16 subjects (3 time points: T0, T1, T2): 8 subjects belong to the group placebo, and the other 8 to the group treatment. So we can consider T0 as an internal control and the placebo group as an external one. I have read the 3.5 section of edgeR manual and taking that example in consideration I design a matrix as followed. If I use the coefficient "TREAT.T1" in the differential expression analysis (LRT) I will obtained the genes that are differentially expressed comparing "TREAT.T1" vs subject (TREAT.T0), is it right? and the same for "TREAT.T2", "CTRL.T1", "CTRL.T2"... If I set "subject" as an intercept I will compare each T1/T2 condition with the corrisponding T0 that in my case is not explicitly defined because it corresponds to the "subject". So not only the first subject (SPI_10) will become the intercept, but all of them according to the treatment, is it right? Thank you in advance for your attention Greta

Yes, your design matrix is correct. The same approach works for any number of repeated measures.

Yes, the TREAT.T1 coefficient is measuring the T1 vs T0 log-fold-change for all treated subjects.