Issue with EDAM ontologies file extension

[toniher]

Description of the bug

I am checking EDAM ontologies automatic assignment loading process here:

tools/nf_core/modules/modules_utils.py

Line 104 in 3ee0bea

def load_edam():

Now, extension is retrieved and processed from the second [1] column of EDAM.tsv. I would rather use an intended field for that, the 15th [14]. With the present assignment, there are some file extensions that are not used in real life (e.g., such as uniprotkb). As a drawback, there may be extensions appearing several times, e.g., json:

json,http://edamontology.org/format_3464
json,http://edamontology.org/format_3969
json,http://edamontology.org/format_3970

In these cases, I would favour assigning the most generic one, that I would hope is the first one is first created in the dictionary (so no overwriting would be allowed).
This will lead to less automatic assignation of filetypes, but more precise ones. Users would need to be recommended to curate manually when more specific EDAM mapping might exist for some specific file extensions such as JSON (e.g., http://edamontology.org/format_3970)

Command used and terminal output

System information

No response

[toniher]

It's worth noting that even though we removed many false positives (e.g., uniprotkb file extension), with this change we now miss some useful ones (such as gff or bed). My take would be check how we could add these file extensions upstream in https://github.com/edamontology/edamontology/ (a debate maybe for another issue could also accommodate different EDAM releases - now latest)

[matuskalas]

Thanks, merged upstream in EDAM. Please post more if needed.

Just one warning: the EDAM.tsv is not (at all) up to date with the EDAM.owl, as we're currently lacking that part of the process. Please let us know in case that this is a showstopper for you, and we can consider prioritising it. Otherwise please let us know if/when you need the http://edamontology.org/EDAM.owl updated with these file extensions (it's not a CD after each merge to main)

[toniher]

Thanks @matuskalas ! The tool used during the linting process provides additions of EDAM links to the stated file formats in the module specifications (detail: https://nf-co.re/blog/2025/modules-ontology). Contributors need to review the additions anyway, but since many of them might not be familiar with EDAM, any initial automation always helps. So, nothing critical, but having a more up-to-date TSV will be beneficial and time-saving. Thanks for the prompt response, and let us know if anything we could help as well.