I have human samples coming from 3 conditions (control, breast cancer and endometrial cancer). I want to do pairwise DE analysis between conditions using EdgeR but i'm not sure how to set up the samples table. Should I use a table with counts from all conditions and compare the conditions of my interest specifying the pair in the exact test function or set up a different table each time with the counts from the conditions i want to compare? Ive tried both ways and they seem to give me different lists of DE genes so I was wondering which one is the correct way to do it. My guess is that the differences come from the estimation of dispersion for each gene.
Sorry if my question is a bit basic but im a beginner.
Thanks a lot for any comments or input you might have.