Fixes all duplicate scoped synonym

[anitacaron]

Two cases were fixed, when:

1. same term label and any other scoped synonym label
2. same label in two different scoped synonym

The cases were fixed to have oboInOwl:hasExactSynonym.

The following sparql update query were used (here it's mixed the fixes for the two cases):

PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX oboInOwl: <http://www.geneontology.org/formats/oboInOwl#>

DELETE {
 ?cls ?synonym ?value .
    ?x a owl:Axiom ;
           owl:annotatedSource ?cls ;
           owl:annotatedProperty ?synonym ;
           owl:annotatedTarget ?value ;
           ?p ?z .
}
INSERT {
 ?cls oboInOwl:hasExactSynonym ?value .
    ?x a owl:Axiom ;
           owl:annotatedSource ?cls ;
           owl:annotatedProperty oboInOwl:hasExactSynonym ;
           owl:annotatedTarget ?value ;
           ?p ?z .
}
WHERE {
  VALUES ?synonym {oboInOwl:hasExactSynonym oboInOwl:hasRelatedSynonym oboInOwl:hasNarrowSynonym oboInOwl:hasBroadSynonym} 
VALUES ?synonym2 {oboInOwl:hasExactSynonym oboInOwl:hasRelatedSynonym oboInOwl:hasNarrowSynonym oboInOwl:hasBroadSynonym} 
  ?cls rdfs:label ?label .
  ?cls ?synonym ?value .
  ?cls ?synonym2 ?value .
  OPTIONAL {
    ?x a owl:Axiom ;
           owl:annotatedSource ?cls ;
           owl:annotatedProperty ?synonym ;
           owl:annotatedTarget ?value ;
           ?p ?z .
  }
  FILTER (str(?label)=str(?value))
  FILTER(?synonym != ?synonym2)
  FILTER( !isBlank(?cls) && regex(str(?cls), "^http://purl.obolibrary.org/obo/UBERON_"))
}

Fixes #2305

[shawntanzk]

converting to a draft so it doesn't accidentally get merged - this needs to be reviewed before merging. I'll post a diff here to make reviewing easier.

[matentzn]

@anitacaron can you check in the update queries(src/sparql/update)? We should do that to make it easier to do similar things in the future. Also, I think we should probably split both problems into separate update queries..

I think you don't capture case 1 entirely because you only look at terms that have at least two synonyms?

[anitacaron]

@matentzn I used two queries to make the changes. I put here the two mixed because the difference is a few lines.

[shawntanzk]

ROBOT diff: https://drive.google.com/file/d/1T9Zt02NLEFUhOaHYNcK7MeInFNk6xVzy/view?usp=sharing

[shawntanzk]

Notice that this adds an additional has_exact_synoym. Can it instead just remove it but move the dbxref annotation to the exact synonym (not sure how difficult that is)

[matentzn]

@shawntanzk dont worry about that, we can use normalisation for contracting these duplicate synonyms! No need to change the SPARQL I think.

[matentzn]

Great job! Assuming this is done?

[shawntanzk]

Assuming this is done?

In the call, there was an agreement that this needs to be reviewed, cant rmbr by who - @dosumis @addiehl? will also look through it tmr :)

[matentzn]

I think the only thing that can be reviewed is wether the synonyms are all exact. Many are not. But this is not really a hill you want to die on here. It will take dozens of curation hours to fix something that has been like this for 15 years, and no one complained so far. I would not get bogged down by details and just focus on QC and infrastructure so we can actually ensure that stuff that is added moving forward is not broken.

[matentzn]

@shawntanzk can we merge this? Seems a waste to let this open for people to never review this..

[shawntanzk]

honestly not super sure, would rather bring this up in a call - people had quite strong opinions about this last time it was discussed

[matentzn]

• Blocked by: Should synonym scopes be contextual? #2363