Humann database errors (a short novel)

[franzosa]

There’s a lot going on here. From the first part, it looks like you’re using a very old HUMAnN (v3.6). HUMAnN 3.5-3.9 were released for compatibility with different versions of MetaPhlAn 4, and they have to be paired with the right version. If you want to stick with HUMAnN 3, I would use v3.9 and install alongside MetaPhlAn following these release notes:

Announcing HUMAnN 3.9

Summary: HUMAnN 3.9 enables tiered search using MetaPhlAn 4.1’s recently updated marker database (June 2023 version) and the bioBakery 3.1 pangenome catalog. HUMAnN 3.9 is a software update only. If you are already using the most recent HUMAnN release (v3.1), there is no need to update your HUMAnN databases to use this version of the software. HUMAnN 3.9 remains backward compatible with MetaPhlAn 3.1 taxonomic profiles (including MetaPhlAn 4.1 run with the v3.1 marker gene database). H…

The other option is to upgrade to HUMAnN 4.0 alpha:

Announcing HUMAnN 4.0 (alpha)

Dear HUMAnN users: HUMAnN 4.0 (alpha) has arrived! Many thanks for your patience—we know there have been a lot of users looking forward to this release for a long time. Since a lot is changing in this version, we’re trying something new and using a Google Doc to organize the release notes (including installation instructions). Please CLICK HERE to check them out. If you have a question about the release notes (e.g. something isn’t clear), then please reply to this post and I will clarify here a…

Which is the latest release and works with MetaPhlAn 4 using its mpa_vOct22_CHOCOPhlAnSGB_202403 index.

In general, when debugging MetaPhlAn-HUMAnN communication, it’s helpful to make sure you can analyze your sample / a demo file with MetaPhlAn outside of HUMAnN, as this can sometimes reveal installation / runtime issues that are not related to the HUMAnN run itself.