I have a conceptual question on microbial functions derived from 16S sequences.

Tools like PICRUSt2 predicts functional potential from 16S sequences. Here, by function, it mostly refers to KEGG orthologs (KOs). Now, if one were to have paired metabolomics data from stool (same as 16s microbiome data), could one consider, at least partially, as some sort of the "truth"? That is, the metabolites that we actually observed and measured should in some way be the gene products, right? Of course, not all metabolites in the stool are "microbiome governed". But could one would expect some sort of concordance between the predictions (KO) from the prediction tools like PICRUSt2 and the measured metabolites from metabolomics data?

I would like to know your thoughts on this. Thank you!

PS: I am thinking of going from KOs to KEGG Compounds via KEGG Reactions.