The bioBakery help forum

Discrepancy between community profile and gene families output


I’m running humann3 on paired metagenomes and metatranscriptomes.
I noticed that I have metatranscriptome reads mapping to species that don’t show up in the community profile output.

e.g. there are metatranscriptome reads mapped to Bacteroides fragilis CAG 47, but Bacteroides fragilis CAG 47 does not show up in the corresponding community profile.
When looking at the metaphlan output it seems that Bacteroides fragilis CAG 47 has been grouped together with Bacteroides fragilis. But this is not the case for the gene families output from both the metagenomes and the metatranscriptomes, where these two strains are separate.
And I see the same for multiple species/strains.

Consequently, I cannot reconcile these species/strains transcription information with their relative abundance.

I use the following commands to process the raw reads:

humann3 -i mg.reads.fastq --protein-database path/to/uniref/ --metaphlan /path/to/metaphlan/3.0.1 -o mg.file --threads 4

humann3 -i mt.reads.fastq --protein-database path/to/uniref/ --metaphlan /path/to/metaphlan/3.0.1 --taxonomic-profile path/to/corresponding_metaphlan_bugs_list.tsv -o mt.file --threads 4

Am I misunderstanding how to process paired metagenomes and metatranscriptomes with humann3?

Thank you for your help!


This looks like our recommended approach to paired mgx + mtx analysis (i.e. profile the mgx sample and use its taxonomic profile to guide the mtx analysis). To confirm, is the file path/to/corresponding_metaphlan_bugs_list.tsv being generated by your first command? I.e. its the taxonomic profile of the corresponding metagenome?

Yes exactly, I use the output from the first command (processing the metagenome sample) as taxonomic profile for the second command (processing the metatranscriptome sample).

OK, that was my guess but I wanted to be sure. :slight_smile: Can you zip up and message me (via Discourse) the mgx taxonomic profile, mgx + mtx genefamilies.tsv files, and mgx + mtx log files? I will take a closer look and then reply here so others can see it.

1 Like

I encounterd the same case, is there any solutions? Even I transfer the metaphlan2 output directly to humann2, the same results were obtained compared to the results without the metaphlan2 output. Thanks in advance.

This ended up being a HUMAnN 3-specific issue (that was improved in the latest release, 3.0.0). If you’re encountering something similar with HUMAnN 2 I would start a separate post with a complete description of the problem (if you haven’t already).

That’s what I have reported Taxonomic results from Humann2 did not consistent with metaphlan2. Thanks for your explanation and useful advisement.