Metaphlan 4 using with Humann3.9

Microbial community profiling HUMAnN
1

Hello everyone,

I’m running into a version compatibility error when trying to use a taxonomic profile from the latest version of MetaPhlAn 4.2.2 as input for HUMAnN 3.9. I would appreciate any advice on the best practice for this workflow.

My Workflow

First, I successfully generated a taxonomic profile using MetaPhlAn 4 with the latest database:

Bash

metaphlan /mnt/d/leopardWGS/host_removed_alignment_results/02_M1_10K_combined.fastq.gz \
    --input_type fastq \
    --db_dir /mnt/d/humann_databases/chocophlan \
    -x mpa_vJan25_CHOCOPhlAnSGB_202503 \
    --nproc 6 \
    -o /mnt/d/leopardWGS/humann_output_M1_10K/taxonomic_profile.txt

Next, I tried to use this profile for functional analysis with HUMAnN 3.9:

Bash

humann --input /mnt/d/leopardWGS/host_removed_alignment_results/02_M1_10K_combined.fastq.gz \
    --output /mnt/d/leopardWGS/humann_output_M1_10K \
    --taxonomic-profile /mnt/d/leopardWGS/humann_output_M1_10K/taxonomic_profile.txt \
    --threads 6 \
    --protein-database /mnt/d/humann_databases/uniref/uniref90_201901b

The Error

This results in the following error, indicating a version mismatch:

ERROR: The MetaPhlAn taxonomic profile provided was not generated with the expected database version. Please update your version of MetaPhlAn to at least v3.0.

My Questions

  1. Is there a way to make the MetaPhlAn 4.x version profile compatible with HUMAnN 3.9, or is this combination not supported?

  2. If not, is the recommended approach to downgrade and use a MetaPhlAn 3 database (mpa_v30_CHOCOPhlAn_201901) to maintain compatibility with HUMAnN 3.9?

  3. For those using MetaPhlAn 4.x version for taxonomic profiling, what is the current best practice for the subsequent functional analysis? Are you using the HUMAnN 4 alpha version?

Thank you for your help!

2

The recent HUMAnN 3.X releases were each made to be compatible with the most recent MetaPhlAn 3 marker database or a specific MetaPhlAn 4 database. HUMAnN 3.9 was the most recent such compatibility release and was designed for the MetaPhlAn 4 Jun23 markers, which are now one step behind the most recent markers (Jan25). We don’t currently have a plan to keep issuing HUMAnN 3.X compatibility releases.

If you’re going for backward compatibility with existing profiles, I would use MetaPhlAn 3 + HUMAnN 3. Otherwise I’d suggest switching to MetaPhlAn 4 + HUMAnN 4 (alpha), with the alpha currently based on the Oct22 markers (with Jan25 in development). Note that the most recent MetaPhlAn 4.2+ is not compatible with the alpha, but we are working on that.

3

I’m facing the same version conflict. I’d like to know if it’s valid to generate taxonomy with MetaPhlAn 4.2.2, since it comes with the updated marker database and use it for diversity analyses such as alpha and beta diversity, and then set up a separate environment with HUMAnN 3.9 and MetaPhlAn 3.1/4.1 for functional profiling???

4

Hello, requesting some insights..

6

Are you considering the below strategy ?

  1. Environment A (The Latest Taxonomy): Running the newest MetaPhlAn (4.2.2+) to get the most accurate, up-to-date taxonomic resolution and using that output for all your alpha/beta diversity analyses and abundance bar plots.

  2. Environment B (The Functional Silo): Running HUMAnN 3.9 tied to an older, compatible MetaPhlAn 4 (with the Jun23 markers) strictly to generate the gene family and pathway abundance files.

If you use this strategy, the taxonomic profiles that drive your functional pathways in Environment B will be slightly different from the standalone taxonomic profiles you generated in Environment A. When writing your methodology for a manuscript, you simply have to be transparent about this. You would state that taxonomy and diversity metrics were calculated using MetaPhlAn v4.2.2 (database Jan25), while functional potential was profiled using HUMAnN v3.9 and its corresponding MetaPhlAn marker database (Jun23). Other parts, MetaPhlAn 4. as the developer (franzosa) mentioned, HUMAnN 3.9 is compatible with MetaPhlAn 4, but only with a specific, older database (the Jun23 markers). It is completely incompatible with the newest MetaPhlAn 4 versions (like 4.2.2 with the Jan25 markers). Because the Biobakery team maps the functional databases (ChocoPhlAn) directly to the taxonomic markers. This is my opinion

1 Like
7

Thank you for sharing your insights. I was exactly considering the strategy you suggested, and I agree that being transparent in the methodology section of the manuscript is important. For the time being, however, I plan to continue using the compatible versions of MetaPhlAn 4.1 and HUMAnN 3.9 with the 2023 database, in order to avoid discrepancies in both taxonomy and functional profiles.

I appreciate your time and guidance