Add CHEBI:is_a as an exact mapping for biolink:subclass_of

[vdancik]

Is your feature request related to a problem? Please describe. We are mapping ChEBI hierarchy to Biolink

Describe the solution you'd like Add CHEBI:is_a as an exact mapping for biolink:subclass_of

What working group (or team) did this request originate from? MolePro Note: This is relevant for members of NCATS Translator.

[balhoff]

Are you using a particular OBO format parser that creates this term? The issue is that there isn't really a term CHEBI:is_a. There is an is_a tag which is a built-in part of the OBO file format, not a particular ontology. The OBO spec defines this as creating an rdfs:subClassOf relationship (or rdfs:subPropertyOf if it is a relation). Maybe you could use rdfs:subClassOf at the point of ingest, which is already mapped to biolink:subclass_of.

[cmungall]

Jim is correct

On Mon, Oct 3, 2022 at 11:03 AM Jim Balhoff @.***> wrote:

Are you using a particular OBO format parser that creates this term? The issue is that there isn't really a term CHEBI:is_a. There is an is_a tag which is a built-in part of the OBO file format, not a particular ontology. The OBO spec https://owlcollab.github.io/oboformat/doc/obo-syntax.html defines this as creating an rdfs:subClassOf relationship (or rdfs:subPropertyOf if it is a relation). Maybe you could use rdfs:subClassOf at the point of ingest, which is already mapped to biolink:subclass_of.

— Reply to this email directly, view it on GitHub https://github.com/biolink/biolink-model/issues/1097#issuecomment-1265602897, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAMMOIS4UCX7NLPL4OBKE3WBLYT3ANCNFSM6AAAAAAQ2AYVZI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[sierra-moxon]

As I understand it, MolePro does want to use biolink:subclass_of in their transformed KG, but they encounter chebi:is_a in their ingest. So they want a chebi:is_a mapping in biolink so their code can automatically know to pick biolink:subclass_of.

[codewarrior2000]

Thanks @sierra-moxon ! This is a page from the ChEBI user manual regarding chebi:is_a https://docs.google.com/document/d/1_w-DwBdCCOh1gMeeP6yqGzcnkpbHYOa3AGSODe5epcg/edit


[balhoff]

In that doc I only see is_a, not chebi:is_a. I think this is an issue to be dealt with in the code parsing the CHEBI OBO file.

[codewarrior2000]

OK, so, CHEBI:is_substituent_group_from is not in the doc but yet it is listed as a narrow mapping for biolink:part_of https://biolink.github.io/biolink-model/docs/part_of.html#relation-part_of. What was the reason for that?

[cmungall]

The uri is incorrect but chebi does indeed declare its own local object property for that

On Mon, Oct 3, 2022 at 1:44 PM Larry Chung @.***> wrote:

OK, so, CHEBI:is_substituent_group_from is not in the doc but yet it is listed as a narrow mapping for biolink:part_of https://biolink.github.io/biolink-model/docs/part_of.html#relation-part_of. What was the reason for that?

— Reply to this email directly, view it on GitHub https://github.com/biolink/biolink-model/issues/1097#issuecomment-1265811415, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAMMOKF2W2BBQXBWGS5TMDWBMLQVANCNFSM6AAAAAAQ2AYVZI . You are receiving this because you commented.Message ID: @.***>

[balhoff]

In the file you will see usages like this:

[Term]
id: CHEBI:58957
name: carboxylatoacetyl group
subset: 3_STAR
def: "The substituent group formed from malonate(1-) ion." []
property_value: http://purl.obolibrary.org/obo/chebi/mass "86.04620" xsd:string
property_value: http://purl.obolibrary.org/obo/chebi/formula "C3H2O3" xsd:string
property_value: http://purl.obolibrary.org/obo/chebi/charge "-1" xsd:string
property_value: http://purl.obolibrary.org/obo/chebi/monoisotopicmass "86.00039" xsd:string
property_value: http://purl.obolibrary.org/obo/chebi/smiles "C(=O)([O-])CC(=O)*" xsd:string
is_a: CHEBI:27207
relationship: is_substituent_group_from CHEBI:30795
is_a: CHEBI:64775

Note how is_substituent_group_from is a value for the relationship tag. This is a property defined in the ontology, not a built-in part of the ontology language. But is_a is part of the file format. The is_substituent_group_from is defined lower down:

[Typedef]
id: is_substituent_group_from
name: is substituent group from
is_cyclic: false
is_transitive: false

You've uncovered another problem in that CHEBI isn't giving the relation a proper identifier, so if we do the expansion for CHEBI:is_substituent_group_from we won't get the same thing that the OBO to OWL mapping gives. :-(

I don't mean to be annoying about it; I just want to make sure we don't fill the Biolink file with IDs that only exist due to parsing issues, rather than actual mappings between different nomenclatures.