Hello, sir.

I have a question regarding using microbiome dataset for DESeq2. I used DADA2 and got like 200 samples X 3000 RSVs table. I want to determine which RSV are differentially abundant across groups.

My question is that should we feed raw 3000 RSVs  to DESeq2, or should we feed agglomerated RSVs (like 100~ genera by phyloseq function tax_glom()) to DESeq2? I read the vignette and find that we are to feed raw count data to DESeq2, but I have many output that differentially abundant across groups which share same genera.

Thanks in advance for help.

Performing differential abundance testing is valid at the ASV level, or at the genus level. The optimal choice really depends on your scientific question, and the phylogenetic depth at which the phenotypic traits relevant to that question are conserved.

This is anecdotal, but when I am doing differential abundance testing in microbiome data, I usually (1) perform independent filtering and remove low prevalence ASVs to reduce the multiple testing burden, (2) test at the ASV level, (3) test at the genus (or somtimes higher) level. So, "and" instead of "or".