Wikidata integration

[dosumis]

EPIC ticket with some suggestions for better wikidata integration. This should be discussed on a future CL call where we agree which items in this epic should be turned into separate tickets. This issue can be closed once all items have been discussed and we've decided which to implement.

Currently mapped terms with Wikipedia links, as known to wikidata: https://w.wiki/6WGX
Currently mapped terms regardless of Wikipedia links: https://w.wiki/7x$K

• Use wikidata mappings to add instant multi-lang support to CL. #1944
• Write SOP for adding mappings to wikipedia directly to wikidata #1945
• Derive wikipedia/wikidata xrefs for CL release file directly from mappings on wikidata #1946

[lubianat]

Some Wikidata constraint checks in CL mappings:

• one CL term to many WD entities: https://w.wiki/6eaQ
• one WD entity to many CL terms: https://w.wiki/6eaT

[github-actions]

This issue has not seen any activity in the past 6 months; it will be closed automatically in one year from now if no action is taken.

[gouttegd]

I’ll investigate the possibility of using SSSOM for that. If we can somehow get a SSSOM mapping set (either directly from Wikidata or possibly through mapping-commons), we could easily derive a component ready to be imported into CL.

[lubianat]

@gouttegd that sounds very doable; I'll work on deriving this SSSOM mapping set,

[lubianat]

@gouttegd here is the SPARQL query for the minimal SSSOM set: https://w.wiki/7x$C

something like:

subject_id	predicate_id	object_id	match_type
wd:Q116343126	skos:closeMatch	http://purl.obolibrary.org/obo/CL_0000735	Complex
wd:Q116446765	skos:closeMatch	http://purl.obolibrary.org/obo/CL_0000744	Complex

(...)

And here is a Wikipedia-only version of the set: https://w.wiki/7x$S

[lubianat]

I've just found this by @dosumis from ~8 years ago: https://github.com/OBOFoundry/obo_wikidata_mapping

Might be relevant.

[matentzn]

cc @andrawaag FYI

[lubianat]

Comming back here, did we have any updates on this? The (basic) SSSOM mapping to add to CL was done, but I am not sure if we have a pipeline in place to generate ontology changes from that. @gouttegd do you have any thoughts?

[gouttegd]

I had no time to work on that until now, but I should be able to find it now that the whole overhaul of Uberon’s composite-metazoan is done. And the upcoming ODK 1.5 will come with the SSSOM plugin built-in, so deriving a component from the SSSSOM set should be easy.

[lubianat]

Nice! Thank you for the reply, @gouttegd. I'll likely talk about that in Biocuration 2024 in India, good to know we have a plan :)