We recently discovered a critical bug in DIAMOND 0.9.36 (as used by HUMAnN 3.0-3.5) that results in substantial loss of high-quality alignments during the translated phase of HUMAnN’s tiered search. DIAMOND’s developer has since confirmed that this bug was present in DIAMOND 0.9.33-0.9.36 but has been corrected in DIAMOND 2.0+. The bug did not affect our benchmarking of HUMAnN 3 in the bioBakery 3 paper (at that time, HUMAnN 3 used DIAMOND 0.9.24) but more recent analyses may have been affected. HUMAnN 3.6 updates the software to use DIAMOND 2.0.15+ for translated search, thus resolving this issue. All HUMAnN 3 users should upgrade to this version and may wish to consider reprofiling affected samples.
- HUMAnN 3.6 now requires DIAMOND 2.0.15+ for translated alignment.
- This change substantially increases the number of reads that can be mapped to UniRef90 (or UniRef50) during translated search after those reads fail to map to a microbial pangenome.
- This change restores the mapping rate and overall performance of HUMAnN’s translated search as observed in our original benchmarking for the bioBakery 3 paper.
- HUMAnN 3.6 is a software/dependency update only. If you are upgrading from HUMAnN 3.1+ then you will not need to update your HUMAnN databases to use this version of the software (DIAMOND 0.9.36 and DIAMOND 2.0+ use the same database format).
- This change does not affect taxonomic profiling with MetaPhlAn or the assignment of reads to species pangenomes (and thus species-stratified function abundances).
- HUMAnN 3.6 remains compatible with MetaPhlAn 3.0+ and MetaPhlAn 4.0 (you can upgrade to HUMAnN 3.6 without upgrading to MetaPhlAn 4.0).
- See here for instructions on installing HUMAnN 3.6 in a new environment:
$ pip install humann –upgrade(will upgrade the existing HUMAnN software).
- You DO NOT need to re-download the latest (v3.1) pangenome database, the DIAMOND-formatted UniRef90/50 databases, or the accessory mapping files to use HUMAnN 3.6. You can point your HUMAnN 3.6 installation to the locations of these existing files using the