Inferring pICRUST2 results

Hello,

I would like a piece of advice regarding resolving functional composition data from pICRUST2.

I have several environmental treatments for the same parent sample, for which after running q2picrust2 pipeline, I get a list of KO IDs with frequencies for each environmental treatment (sample-ID). I was wondering whether we could treat each treatment as a “separate metagenome”, and say compare frequencies of related KO IDs across the sample-IDs? After reading through pICRUST readme/tutorial files, I don’t see any discussion on this, so I would like to verify if this is scientifically accurate.

Thank you,
Anirban