HUMAnN 3.8 - How to get taxonomic results at the order level instead of genus?

[Jeremy_Tournayre]

Hello,

I’ve been using HUMAnN 3.8.
I got many unclassified results because I think the taxonomic resolution of genus is too precise.
So, I’m interested in obtaining taxonomic results at the order level rather than at the genus level.
Unfortunately it seems that is not possible? Can you please help me ?

Best regards,
Jérémy Tournayre

[MARTIN_MIHULA]

Dear Jeremy,

I’m not sure whether you can specify the taxonomic level in the HUMAnN directly. Nevertheless, you can ran MetaPhlAn separately and specify the level there:

--tax_lev 'o' (orders only)

Then you can ran HUMAnN and refer to the MetaPhlAn output:

--taxonomic-profile metaphlan_output

I hope it solves your problem!

[Jeremy_Tournayre]

Hello,

I found solution here:

Request for Higher Taxonomic Level Analysis in HUMAnN

Dear HUMAnN Development Team, I am currently utilizing the HUMAnN software for metagenomic analysis, specifically interested in taxonomic characterization at various levels of resolution. The output I am receiving is at the species level by default, and I have successfully used the humann_genefamilies_genus_level command to obtain results at the genus level. However, my research requires analysis at higher taxonomic levels beyond genus. Could you please advise if there is a way to modify the o…

with humann_infer_taxonomy

humann_infer_taxonomy -l Genus -d uniref90-tol-lca -i humann_genefamilies.tsv -o Genus/humann_genefamilies.tsv"

It works but I don’t know why there are some stange things like this below.
Why do I have “unclassified” instead of “g__Prevotella” and “g__Prevotellaceae_unclassified” in the Genus file?

Original “humann_genefamilies.tsv” :

UniRef90_A0A010YBU9 43.4636486692
UniRef90_A0A010YBU9|g__Prevotella.s__Prevotella_sp_ne3005 15.4012377598
UniRef90_A0A010YBU9|g__Prevotella.s__Prevotella_ruminicola 12.1212121212
UniRef90_A0A010YBU9|g__Prevotellaceae_unclassified.s__Prevotellaceae_bacterium_MN60 6.0506050605
UniRef90_A0A010YBU9|g__Prevotella.s__Prevotella_sp_tc2_28 4.4004400440
UniRef90_A0A010YBU9|g__Prevotella.s__Prevotella_sp_tf2_5 3.2789872934
UniRef90_A0A010YBU9|g__Prevotella.s__Prevotella_aff_ruminicola_Tc2_24 2.2111663903

In Phylum:

UniRef90_A0A010YBU9 43.4636
UniRef90_A0A010YBU9|p__Bacteroidetes 43.4636

In Genus:

UniRef90_A0A010YBU9 43.4636
UniRef90_A0A010YBU9 unclassified  43.4636

Best regards,
Jérémy Tournayre

[franzosa]

Jérémy’s answer is correct. HUMAnN works with species-level pangenomes, so providing genus-level taxonomy as input will just forward all reads to translated search. Using the infer_taxonomy script you can regroup known species hits to a higher level (e.g. genus) and you can also make guesses about the taxonomy of unclassified gene family hits based on the LCA of the species that contributed to the gene family. We don’t offer a means to directly profile against genera, for example.