Time-series experiment with patients from 2 conditions and 5 time-points from each patient
Hi everyone, I am trying to create a DEseq2 model matrix for my RNA-seq experiment. I have two conditions (treated vs untreated) and in each condition 10 patients (10 different patients in treated, 10 different patients in untreated group) and from each patient I have 5 time-points (baseline,10min, 3h, 6h, 24h). I would like to identify DE genes whose expression profiles across my time-points are different in the treated vs untreated group. I am now struggling with how to create my design for DESeq2. If my time-points were independent (not always 5 from the same patient), I would go for full model
~ condition + time + condition:time and reduced
~ condition + time. But when I ignore the non-independency of my time-points, in the PCA plot I can see that samples cluster more based on the patient they originated from rather than time-points or conditions. So I have a patient batch effect I need to remove. But I am not able to find out how (or if) that is possible to do with DESeq2, because if I add my patient effect into the model (like this
~ patient + condition + time + condition:time) I get an error that my model matrix is not full rank (which makes sense since I do not have the same patients in both groups).
My experimental design looks like this:
sample condition time patient 01_A_SE treated baseline pat01 01_C_SE treated 10min pat01 01_D_SE treated 3h pat01 01_E_SE treated 6h pat01 01_F_SE treated 24h pat01 02_A_SE treated baseline pat02 02_C_SE treated 10min pat02 02_D_SE treated 3h pat02 02_E_SE treated 6h pat02 02_F_SE treated 24h pat02 . . . . . . . . . . . . 18_A_SE untreated baseline pat18 19_C_SE untreated 10min pat18 19_D_SE untreated 3h pat18 19_E_SE untreated 6h pat18 19_F_SE untreated 24h pat18 20_A_SE untreated baseline pat20 20_C_SE untreated 10min pat20 20_D_SE untreated 3h pat20 20_E_SE untreated 6h pat20 20_F_SE untreated 24h pat20
Is there a way how do this in DESeq2, or do I have to look into limma or some other software?
Thank you very much!