Upto what extent functional potential of microbiome predicted by HUMAnN is supported by original functional profile (i.e., transcriptome or metabolome) data?

Hi @franzosa HUMAnN profiling of metagenomic data is largely predictive in nature. Is there any study exist that validates HUMAnN profiles with transcriptome/metabolome data?


I’m not sure what you mean exactly? When I think of “predictive” functional profiling my mind goes to something like PICRUSt (which is trying to guess about the genes in a community based on its taxonomy). HUMAnN is directly quantifying genes from metagenomic sequencing data.

If you’re asking about how much of this functional potential is active, we do see that the majority of gene families are transcribed in a microbiome (in the human gut at least, where we do most of our work). Validating activity based on metabolomics is much harder as the mapping between genes and metabolites is not very comprehensive.

1 Like

Yes, I am asking about those “active” genes which are being transcribed. When they are being transcribed are they following the same/similar relative abundances in the transcription level which is captured from the metagenome level by HUMAnN.


Yes, that does seem to be the case, i.e. the “null model” is that if you have more gene copies (MGX) you tend to see more transcripts (MTX) – as if all genes were transcribed at the same rate. Differential expression in the context of MTX is then looking for more/fewer transcripts than you would’ve expected based on the gene copy number of the function of interest. We discuss these ideas in more detail in this paper: