Hello! I'm trying to run DESeq2 on samples that belong to different treatment groups. As an example:

                treatment_groups

Sample 1 responder to drug A

Sample 2 responder to drug A

Sample 3 non responder to drug A

Sample 4 non responder to drug A

Sample 5 responder to drug B

Sample 6 responder to drug B

Sample 7 non responder to drug B

Sample 8 non responder to drug B

I tried running DESeq2 on all the samples together and use results(dds, contrast=c("treatment_groups", "responder to drug A", "non responder to drug A")) to get the differential expression results for drug A response. I also tried just separating the samples and only have sample 1-4, which produced a somewhat different differential expression results. Is it recommended to do this one way or the other? Thanks in advance.

In general, include all the samples you've got, to get better dispersion estimates and size normalization.