Closed kevinschaper closed 2 years ago
I don't see anything problematic in the predicate column of the tsv
$ cut -f 3 monarch-kg_edges.tsv | sort | uniq -c | sort -rn
1376757 biolink:category
994966 biolink:expressed_in
933102 biolink:subclass_of
484133 biolink:actively_involved_in
428297 biolink:has_phenotype
420569 biolink:enables
365084 biolink:orthologous_to
321122 biolink:located_in
168019 biolink:acts_upstream_of_or_within
97949 biolink:active_in
85398 biolink:participates_in
80481 biolink:part_of
18696 biolink:related_to
9902 biolink:has_attribute
4591 biolink:contributes_to
4221 biolink:develops_from
3652 biolink:regulates
3196 biolink:has_output
3089 biolink:in_taxon
3088 biolink:negatively_regulates
3073 biolink:positively_regulates
2958 biolink:colocalizes_with
2645 biolink:has_input
2541 biolink:has_part
2534 biolink:overlaps
2160 biolink:temporally_related_to
1783 biolink:has_participant
1342 biolink:affects_transport_of
1178 biolink:acts_upstream_of
1125 biolink:associated_with
998 biolink:coexists_with
867 biolink:subPropertyOf
812 biolink:capable_of
701 biolink:interacts_with
469 biolink:occurs_in
406 biolink:causes
391 biolink:acts_upstream_of_or_within_positive_effect
349 biolink:affects
305 biolink:preceded_by
174 biolink:caused_by
169 biolink:acts_upstream_of_or_within_negative_effect
155 biolink:inverseOf
152 biolink:acts_upstream_of_positive_effect
121 biolink:acts_upstream_of_negative_effect
104 biolink:disrupts
100 biolink:expresses
52 biolink:produced_by
35 biolink:type
33 biolink:produces
31 biolink:derives_from
30 biolink:precedes
12 biolink:homologous_to
10 biolink:location_of
9 biolink:increases_degradation_of
6 biolink:has_variant_part
6 biolink:affects_localization_of
5 biolink:has_unit
1 predicate
1 biolink:object
1 biolink:model_of
1 biolink:correlated_with
oh! I see it!
biolink:subPropertyOf
In the super short term, I think I might try to just patch this with sed to biolink:sub_property_of
- which isn't a real biolink predicate, but would pass the validation.
@sierra-moxon it looks like this is coming from the kgx conversion of monarch ontology, could it just be that we're not up to date with kgx, or do you think it's something else to dig into on the ontology->kgx conversion?
doh, also biolink:inverseOf
This is coming up downstream when running Plater against Neo4j loaded from monarch-kg.tar.gz, and the problem might actually be with the merge, but it looks like we have a bad predicate value showing up