Difference between DESEq2 and ANCOM for estimating biomarkers (which is preferrred?)

Question

[eimanpharmacist]

Hello, I am doing my microbiome analysis for the first time. To estimate biomarkers that separate healthy group from disease one, I used DESEq2 and ANCOM with the same filtration criteria (remove any taxa with read count <10). I know each has a different algorithm, and I found an overlap between them in the identified biomarkers, whereas other indicator species were exclusively identified from each pluging. I don't know if is it accepted to mention both results in the manuscript and which one should I stick? Thanks Eman

Answer 1

[Charles Warden]

I haven't personally tested ANCOM, but I've tested MetagenomeSeq.

I think it would be safe to use an overlap between methods, but you may find that too conservative in some cases.

If you have some way to justify a particular choice (to match your overall feel for results, or validate a known difference between samples), I think it is OK to have some subset of work that you represent in your paper (if allowed, I would think of bench-marking as something to have a supplemental data, to help the reader focus on particular messages in the main text).