Hi everyone,
When performing differential abundance analysis using DESeq2, should we use the unrarefied (raw) feature table, or should the feature table first be rarefied to the same sequencing depth used for alpha and beta diversity analyses?
1 Like
Hello @asmaamorsi !
You should always use the raw feature table when running DESeq2. From the DESeq2 paper:
Note that while the rlog transformation builds upon on our LFC shrinkage approach, it is distinct from and not part of the statistical inference procedure for differential expression analysis described above, which employs the raw counts, not transformed data.
Out of curiosity, is there a specific reason you chose DESeq2 instead of other methods like ANCOM-BC? Some years ago I came across Gloor et al., 2017, which really made me rethink how to handle microbiome data 
Hope this helps!
Best,
Sergio
4 Likes