Capture exact, related, narrow, broad synonyms in specific fields

[kevinschaper]

In the Monarch UI, we'd like to distinguish between synonym types (maybe show on hover). I think my ideal would be to continue populating synonyms as-is (possibly all synonyms?), while also capturing exact synonyms in exact_synonym, broad in broad_synonym, etc

[caufieldjh]

This is another area where the KGX format diverts from what Biolink specifies (and maybe another argument for a unified kgx-Biolink schema).

Not a technical limitation by any means, but a meaningful distinction to keep in mind for other ingests and graph transforms.

[matentzn]

😱 How is this possible - I though KGX was specifically made to be the TSV file format for biolink? How exactly is it different here?

[caufieldjh]

KGX was made to be the format for Biolink and other non-Biolink KGs, so it's not a complete match in all ways.
In this case, the default KGX format has just one synonym field, synonym.
There isn't anything in it explicitly forbidding splitting synonyms by type and even in Biolink the four synonym subtypes are all is_a: synonym:
https://github.com/biolink/biolink-model/blob/c16ad903282be7733676788b75604580f03bf602/biolink-model.yaml#L594-L635

[caufieldjh]

loosely related: biolink/kgx#218

[caufieldjh]

I think this may need to be addressed in kgx as there's data loss when phenio.json is transformed to KGX TSV; the former has synonym types:

      "id" : "http://purl.obolibrary.org/obo/CHEBI_17736",
      "lbl" : "S-(4-bromophenyl)-L-cysteine",
      "type" : "CLASS",
      "meta" : {
        "subsets" : [ "http://purl.obolibrary.org/obo/chebi#3_STAR" ],
        "synonyms" : [ {
          "synonymType" : "http://purl.obolibrary.org/obo/chebi#IUPAC_NAME",
          "pred" : "hasExactSynonym",
          "val" : "(2R)-2-amino-3-[(4-bromophenyl)sulfanyl]propanoic acid",
          "xrefs" : [ "IUPAC" ]
        }, {
          "pred" : "hasExactSynonym",
          "val" : "S-(4-Bromophenyl)-L-cysteine",
          "xrefs" : [ "KEGG_COMPOUND" ]
        }, {
          "synonymType" : "http://purl.obolibrary.org/obo/chebi#IUPAC_NAME",
          "pred" : "hasExactSynonym",
          "val" : "S-(4-bromophenyl)-L-cysteine",
          "xrefs" : [ "IUPAC" ]
        } ],
...

but the latter does not:

~/kg-phenio$ head -1 data/transformed/phenio/phenio_node_sources_nodes.tsv && grep CHEBI:17736 data/transformed/phenio/phenio_node_sources_nodes.tsv 
id      category        name    description     xref    provided_by     synonym deprecated      iri     same_as subsets
CHEBI:17736     biolink:ChemicalEntity  S-(4-bromophenyl)-L-cysteine            Beilstein:3204780|KEGG:C03900   infores:chebi   (2R)-2-amino-3-[(4-bromophenyl)sulfanyl]propanoic acid|S-(4-Bromophenyl)-L-cysteine|S-(4-bromophenyl)-L-cysteine                http://purl.obolibrary.org/obo/CHEBI_17736              3_STAR