OMIM synonym scope needs review

[nicolevasilevsky]

Action items:

• [ ] 1. most OMIM synonyms should be exact (not related)
• [ ] 2. most NCIT and GARD syns should be exact too

I think we could do this:

1. change all OMIM synonyms from related to exact,
2. then check where there are duplicate synonyms coming from OMIM and manually review these, and then make them related synonyms, if applicable (or maybe broad, in the example from Donna below). (it would probably be good to do this for all synonyms).
3. (From @maglott) We should do a special review of alternate names for 'disease name 1' because it may be that some of the alternate names are more correctly synonyms for 'disease name'

Related to https://github.com/monarch-initiative/mondo/issues/1794

I just created this and realized it is a duplicate of https://github.com/monarch-initiative/mondo/issues/2255

[nicolevasilevsky]

From @maglott on #1794 (Just copying over the comments)

I am a bit skeptical about 'All OMIM synonyms should be exact (not related)'

• there are cases like https://www.omim.org/entry/602429 and https://www.omim.org/entry/603383 where the same alternate name is used on multiple records.
• there are cases like the records for gene-specific name 1 where one of the alternate name is likely synonymous with the phenotypic series, not the gene-specific entity.

Are these cases being detected?

[joeflack4]

@nicolevasilevsky Do you think you could take a look at the output file of the OMIM ingest and see if it is satisfactory? Regarding action item (1), I looked and there are ~414~ ~48,974 exactSynonyms and ~158~ 18,358 related synonyms in the output. For action item (2), I don't see 'GARD' or 'NCIT' in the prefix list, so not sure how to check that.

[matentzn]

@nicolevasilevsky please bring this issue to the next QC call, as this requires discussion.

[nicolevasilevsky]

we can make a blanket assumption that related OMIM syns should be exact but it will need manual review

run query - in branch, we'll see diff and we can click on change in GitHub Desktop and can undo in GH Desktop

[maglott]

I still think they should be exact only if unique.

[matentzn]

I agree @maglott - @nicolevasilevsky we can easily test for uniqueness! Just make sure to remind me when we work on this!

[nicolevasilevsky]

Action item:

• [ ] spot check the ttl file (3 classes) and compare to the OMIM record and make sure the exact syns are alternative terms and included terms are related
• [ ] ask Nico and joe to update the existing Mondo terms