HUMAnN/MetaPhlAn error

Hi @Maria_Fernandes-Mar

Thanks for explaining!

Since MetaPhlan wasn’t classifying your reads well anyway (the newest version, when used directly?) you could try using the Kraken2 output instead in HUMAnN with this converter tool.

  • Krakentools: Convert kraken report file to MetaPhlAn-style

To be clear: it looks to me like the updated indexes are included in MetaPhlAn at UseGalaxy.eu (last updated March 2025). This result can be passed through to HUMAnN. The current usability problems are:

  1. the indexes at EU need to be added a specific shared index resource to make them available at other UseGalaxy servers → Add updated Metaphlan index(es) to CVMFS · Issue #52 · galaxyproject/idc · GitHub (original asked at Metaphlan vJun23)
  2. the file labeling is inconsistent between servers, leading to workflow/shared data table issues for administrators, and confusion for scientists → Make Metaphlan/Humann DBs transferable · Issue #5681 · galaxyproject/tools-iuc · GitHub
  3. those same indexes could be added to the HUMAnN tool as well (what you originally asked about) but the other two items are delaying the consolidation. These databases are large enough to prevent an easy fix (replicate them with multiple names) and multiple labels leads to reproducibility issues we want to avoid. The available path for now is to run MetaPhlAn directly instead a the server that hosts the current index, and to pass the result to HUMAnN for the additional classification layers.

So, there are a few technical issues remaining to be resolved to make this all nicer (for scientists and admins!), but the scientific path with the most current data should be fine. Adding all the tools to a simple workflow would produce about the same total “processing time” per sample. If you need to move data between servers, this can all be done via URL (no download step).

Apologies for the complicated usage! But, does this help? We can follow up more. :slight_smile: