RTXteam / RTX-KG2

Build system for the RTX-KG2 biomedical knowledge graph, part of the ARAX reasoning system (https://github.com/RTXTeam/RTX)
MIT License
34 stars 9 forks source link

KG2 Validation issue #345

Closed edeutsch closed 10 months ago

edeutsch commented 10 months ago

The following external issue: https://github.com/NCATSTranslator/Feedback/issues/495 shows:

    $ infores:semmeddb -> infores:rtx-kg2 -> infores:arax
        # activity:
        - edge_id | qualifier_type_id: 
            NCBIGene:4763[biolink:Gene|biolink:Protein]--biolink:affects->MESH:D020558[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:5027[biolink:Gene|biolink:Protein]--biolink:affects->MESH:C545445[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:5027[biolink:Gene|biolink:Protein]--biolink:affects->MESH:D015850[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:5027[biolink:Gene|biolink:Protein]--biolink:affects->MESH:D016207[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:5027[biolink:Gene|biolink:Protein]--biolink:affects->MESH:D016328[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:5027[biolink:Gene|biolink:Protein]--biolink:affects->MESH:D053583[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:3351[biolink:Gene|biolink:Protein]--biolink:affects->UMLS:C0036751[biolink:ChemicalEntity] | biolink:object_aspect_qualifier
            NCBIGene:5027[biolink:Gene|biolink:Protein]--biolink:affects->PUBCHEM.COMPOUND:5460341[biolink:ChemicalEntity|biolink:SmallMolecule|biolink:Drug] | biolink:object_aspect_qualifier

I think the issue is that "activity" is an invalid object_aspect_qualifier

ecwood commented 10 months ago

This might resolve with the fix for #339

saramsey commented 10 months ago

@edeutsch I'm seeing the following in biolink-model.yaml. Seems to me that activity is an allowed value for GeneOrGeneProductOrChemicalEntityAspectEnum, which (per private communication with Matt Brush) is the definitive list of allowed values for "aspect qualifiers" in Biolink: https://github.com/biolink/biolink-model/blob/54d9e7ac2023ff97ee634411abdc2a41d359862e/biolink-model.yaml#L10814

Can you help me to understand what makes us think "activity" is not allowed? Or not allowed in this specific context?

saramsey commented 10 months ago

@edeutsch you are right that the aforementioned triple (the first one) is missing a object_direction_qualifier. I concur with @ecwood that this is a known issue already fixed in code and for which the fix should be rolled out in KG2.8.6c.

edeutsch commented 10 months ago

After consultation with Richard, I think I understand now. he said:

This validation error is a known issue and arises because Gene--affects->ChemicalEntity does not have a biolink:Association documenting qualifier ranges (in slot_usage) 4:30 I have noted, however, that an inverse association, ChemicalEntity--affects-->Gene is available 4:30 and does have the require qualifier definitions 4:33 That is the same issue as described as part of my comment yesterday 4:34 (actually, the true inverse association should be ChemicalEntity--affect_by-->Gene

So the way understand it, 'activity" is permitted for Chem affects Gene. But not Gene affects Chem. but this is an acknowledged BL oversight, and will be fixed in BL.

So my understanding is no fix from us is needed here. closing.