Missing edge categories?

[caufieldjh]

@kevinschaper reports that kg-phenio may be missing edge categories.

• Check if edge categories are present in the KGX output.
• If not, investigate the transform and the PHENIO build.
• If so, ensure they are consistent with expectations (e.g., in expected column, etc)

[caufieldjh]

KG-Phenio has edge properties, they're just quite minimal. This is the header:

id	subject	predicate	object	category	relation	knowledge_source

Not much going on there!

For comparison, here's the Monarch graph heading:

id	original_subject	predicate	original_object	category	aggregator_knowledge_source	primary_knowledge_source	publications	qualifiers	provided_by	has_evidence	stage_qualifier	relation	knowledge_source	negated	frequency_qualifier	onset_qualifier	sex_qualifier	evidence	subject	object

Not all properties will be necessary for PHENIO, but the knowledge sources can certainly be expanded.

[caufieldjh]

The merged Upheno mapping table also needs a knowledge_source added, but that can just be added at transform time as a KGX argument.

[caufieldjh]

Also want:

• synonyms and xrefs (see No synonyms coming in from Phenio for Mondo terms monarch-initiative/monarch-ingest#409 and Include xrefs #79)
• obsoletions (Expand node properties #90)
• other provenance details contained in PHENIO but not included in the KG (PMIDs? versions?)

[kevinschaper]

I was just poking at some kgx validation output and I noticed I had complaints about ZFA being an invalid prefix in ZP->ZFA associations, and it might have something to do with the blank categories. Here's a summary of edges with blank categories

category	subject_namespace	predicate	object_namespace	primary_knowledge_source	count(*)
	GO	biolink:subclass_of	GO	infores:go	83592
	ZP	biolink:subclass_of	ZP	infores:zp	60886
	FBbt	biolink:related_to	FBbt	infores:fbbt	52322
	MONDO	biolink:subclass_of	MONDO	infores:mondo	38795
	XPO	biolink:subclass_of	XPO	infores:xpo	35717
	FBbt	biolink:subclass_of	FBbt	infores:fbbt	31665
	UBERON	biolink:subclass_of	UBERON	infores:uberon	23079
	HP	biolink:subclass_of	HP	infores:hp	22532
	GO	biolink:related_to	GO	infores:go	20260
	UBERON	biolink:related_to	UBERON	infores:uberon	19247
	MP	biolink:subclass_of	MP	infores:mp	18334
	WBbt	biolink:subclass_of	WBbt	infores:wbbt	8168
	EMAPA	biolink:related_to	EMAPA	infores:emapa	7037
	WBbt	biolink:related_to	WBbt	infores:wbbt	6924
	CHEBI	biolink:subclass_of	CHEBI	infores:chebi	6559
	ZP	biolink:related_to	ZFA	infores:upheno	5840
	ZP	biolink:related_to	GO	infores:upheno	5822
	CHEBI	biolink:related_to	CHEBI	infores:chebi	5043
	EMAPA	biolink:subclass_of	EMAPA	infores:emapa	4545
	EMAPA	biolink:subclass_of	UBERON	infores:emapa	4477
	WBPhenotype	biolink:subclass_of	WBPhenotype	infores:wbphenotype	3364
	ZFA	biolink:subclass_of	ZFA	infores:zfa	3199
	MONDO	biolink:related_to	UBERON	infores:mondo	3027
	WBbt	biolink:subclass_of	GO	infores:wbbt	2766
	ZFA	biolink:related_to	ZFA	infores:zfa	2752
	MP	biolink:related_to	UBERON	infores:upheno	2577
	MONDO	biolink:related_to	MONDO	infores:mondo	2458
	GO	biolink:related_to	CHEBI	infores:go	2090
	ZFA	biolink:subclass_of	UBERON	infores:zfa	2071
	HP	biolink:related_to	UBERON	infores:upheno	1588
	MONDO	biolink:related_to	HP	infores:mondo	1450
	GO	biolink:related_to	UBERON	infores:go	1124
	MPATH	biolink:subclass_of	MPATH	infores:mpath	946
	MP	biolink:related_to	GO	infores:upheno	870
	FBbt	biolink:related_to	GO	infores:fbbt	571
	ZP	biolink:related_to	CHEBI	infores:upheno	455
	MONDO	biolink:related_to	GO	infores:mondo	432
	UBERON	biolink:related_to	GO	infores:uberon	423
	FBbt	biolink:subclass_of	UBERON	infores:fbbt	369
	HP	biolink:related_to	CHEBI	infores:upheno	359
	WBPhenotype	biolink:related_to	GO	infores:upheno	325
	HP	biolink:related_to	GO	infores:upheno	279
	WBPhenotype	biolink:related_to	WBbt	infores:upheno	264
	MP	biolink:related_to	CHEBI	infores:upheno	191
	XPO	biolink:related_to	GO	infores:upheno	141
	ZP	biolink:related_to	MPATH	infores:upheno	137
	MP	biolink:related_to	MPATH	infores:upheno	134
	HP	biolink:related_to	MPATH	infores:upheno	71
	MONDO	biolink:related_to	CHEBI	infores:mondo	49
	WBbt	biolink:subclass_of	UBERON	infores:wbbt	49
	MP	biolink:related_to	MP	infores:upheno	32
	UBERON	biolink:related_to	CHEBI	infores:uberon	32
	WBPhenotype	biolink:related_to	CHEBI	infores:upheno	21
	MPATH	biolink:related_to	MPATH	infores:mpath	3
	HP	biolink:related_to	HP	infores:upheno	2
	UBERON	biolink:subclass_of	GO	infores:uberon	2
	WBPhenotype	biolink:related_to	UBERON	infores:upheno	1

[cmungall]

I am not sure we have an implemented strategy for populating edge categories when going from owl->kgx

This could be done in kgx by inference,

  gene to phenotypic feature association:
    is_a: association
    exact_mappings:
      - WBVocab:Gene-Phenotype-Association
    defining_slots:
      - subject
      - object
    mixins:
      - entity to phenotypic feature association mixin
      - gene to entity association mixin
    slot_usage:
      subject:
        range: gene or gene product
        description: "gene in which variation is correlated with the phenotypic feature"
        examples:
          - value: HGNC:2197
            description: "COL1A1 (Human)"
      object:
        range: phenotypic feature

however I would do this with linkml:classification_rules now

this will probably not be straightforward to add to kgx - @kevinschaper how much does our validation strategy depend on this being present

[kevinschaper]

It looks like we're hitting some kgx validation issues within translator infrastructure that might be coming in from blank category fields on edges. Would it work to just fill in with biolink:Association rather than nulls?

(I think I might do that in my phenio kgx massaging, and of course it won't do anything once they're set)