Release notes for HUMAnN 3.0.0 and ChocoPhlAn 201901b

Hi All - HUMAnN 3.0.0 has been out in the wild for a little while now, but I was slow in getting these release notes together. Please reply with any questions. Thanks! -Eric


  • HUMAnN 3.0.0 is a major update to HUMAnN (ending the HUMAnN 3.0.0.alpha lineage). Users will need to update their HUMAnN software and all database files for this release. Installation instructions can be found here: humann3 – The Huttenhower Lab.

Database updates

  • The ChocoPhlAn database has been updated with ~600 additional species pangenomes that were missing from the previous release. This and other updated databases are tagged “201901b” to differentiate them from the previous (“201901”) release.

  • The UniRef protein databases have been re-indexed for the latest DIAMOND 0.x format (HUMAnN has not transitioned to DIAMOND 2.0).

  • MetaCyc pathway definitions and related information have been updated to release 24.0.

  • The utility mapping database has been updated with expanded UniRef-to-EC maps, updated UniRef taxonomy maps, and a much-requested UniRef90-to-UniRef50 map.

Performance updates

  • Changed a behavior that caused pangenomes of close relatives of MetaPhlAn-detected species (including many redundant CAG species) to be automatically added to the sample-specific pangenome index.

  • Improved memory efficiency during post-processing of mapping results.

Tool updates

  • The functionality of the humann_infer_taxonomy tool (which assigns putative taxonomy to unclassified UniRef abundance) has been restored and improved using new error-corrected UniRef LCA definitions.

  • The humann_barplot tool has been redesigned to label more community species and to facilitate cross-plot consistency.

  • Several HUMAnN 2.0-era experimental tools have been deprecated.

Quality-of-life changes

  • The HUMAnN command-line interface has been reorganized to group together related arguments (e.g. all arguments related to translated search).

  • Miscellaneous small bug fixes.

Thank you on behalf of the Nephele team. We now have updated the the WGS pipeline we host on Nephele that is using biobakery workflows (0.13.2) and HUMAnN 3.0.

Hi, I am using the latest human(3) and metaphlan with the updated databases. My metagenomes are heavily dominated by fungal reads(70-80%). However, human does not classify most of the reads (90%). I also do not see any pathways associated to fungal reads. I have kept the command quite simple

humann --input $in --output $out --threads 64

Can you help me tweak the comment to maximise the eukaryote read classification?

Thanks for the great tool by the way.

Can you please copy this question to a new post?