Cannot regroup into kegg pathway in Humann3

Hi communities,
I tried to regroup the pathway into kegg pathway by following command.

humann_regroup_table -i merged_path.tsv -o ./path_kegg.tsv -g uniref90_ko

However, there was only unintegrated pathway in the output file.
Even when I tried to regroup into go pathway, there was the same issue.

How can I regroup the pathways with other databases…?


You can regroup from UniRefs (gene families) to KOs (broader gene families), which is what the -g uniref90_ko argument is asking for (but it would require a genefamilies.tsv input file rather than pathways). We don’t currently support regrouping pathways with the built-in mapping options, and it would be challenging to regroup from MetaCyc pathways to KEGG pathways. You would be better off computing KOs from your gene families and then trying to compute KEGG pathways based on the KOs you found.

Thank you so much for replying me.
Do you have any recommend packages or softwares for computing KEGG pathways easily…?
It seems difficult to compute KEGG pathways from gene families one by one.

Thank you,

Hi, I am using humann2 and I wish to regroup HUMAnN2 gene abundance table using a mapping of GO terms. Basically I wish to categorise them in different metabolic process groups. I couldn’t find much online. I tried the following but it gives 0 entries renamed: I also tried go, metacyc_pwy instead of kegg_pathway but nothing seems to work.
humann2_renorm_table -i test_genefamilies.tsv -u relab -p -o test.ra.tsv
humann2_regroup_table -i test.ra.tsv -o test.tsv -g uniref90_rxn -e 6

humann2_rename_table -i test.tsv -o test.ko.tsv -n kegg-pathway

Any help would be appreciated. Thank you in advance

You want to be regrouping to GO (or the Informative GO terms, infogo1000) bundled with HUMAnN in the second command. After that you’ll be able to rename them with GO descriptions in the third command. If you don’t have GO as an option for your regroup_table command you need to install the HUMAnN utility mapping database using the database script.

Hi Eric,
Thank you very much for your quick response. I will try that, however is there a way to group these into specific metabolic processes? For example to group uniref families into cellular amino acid metabolic process, carbohydrate metabolic processes, lipid metabolic processes and xenobiotics metabolic processes.

Thank you

Hi, sorry I don’t think I was able to place the question properly.
Basically, Is there a mapping file for metacyc pathways that could be used for regrouping? And then renaming using metacy-name-pathway file?
Would that be the approach here?
Any help would be appreciated

Thank you in advance

For your first question, the best bet for grouping UniRefs really broadly would probably be to group them to GO terms (using the file we supply) and then “bubble up” more specific GO terms to broader terms based on the GO hierarchy. HUMAnN doesn’t have any embedded tools for doing this, however.

For your second question, we don’t have any options for regrouping MetaCyc pathways into broader categories bundled with HUMAnN. MetaCyc does maintain its own hierarchy with broader pathway categories, and there might be an old version of that file floating around on the forum, but it’s not something we specifically updated for the latest HUMAnN release.

Hi @franzosa , thank you I will try this approach. So basically there is no mapping file to use for regrouping to metacyc? I thought there would be one as there is a metacyc-pathway-name.txt file. Or is there a mapping file which is not part of the utility mapping files folder?
Thank you

Correct, we don’t bundle mapping files for regrouping MetaCyc pathways into other entities. The main idea of regrouping is to collapse large numbers of highly specific UniRef gene families into smaller numbers of broader molecular functions to simplify analysis and interpretation of a dataset (we compute pathways from gene families for the same reason, albeit by other means).

Thank you very much, greatly appreciated.