Hello,
I have a question about the DESeq2 package.
My experiment is comprised of eight different treatments, which I am analyzing in pairs.
I did the DESeq2 analysis twice for each pairwise comparison:
1) The first time only the counts files for the two compared treatments were taken in consideration for both the normalization and the Log2FC analysis.
2) The second time all the counts files from all eight treatments were normalized together and only the two compared treatments were then used for the ddseq function.
The results from those two strategies were a bit different, with pairwise comparisons using the second strategy leading to roughly double the amount of statistically significant differentially expressed genes than when using the first strategy.
I would appreciate if you could explain to me the basis for this difference. Should I prefer one of these approaches over the other?
Thanks a lot,
