Closed caufieldjh closed 3 months ago
Stats for the 20240313 KG build say that only 3 OMIM nodes are there, and they are these:
OMIM:has_inheritance_type biolink:has_attribute Graph http://purl.obolibrary.org/obo/OMIM_has_inheritance_type
OMIM:has_manifestation biolink:superclass_of Graph http://purl.obolibrary.org/obo/OMIM_has_manifestation
OMIM:manifestation_of biolink:manifestation_of Graph http://purl.obolibrary.org/obo/OMIM_manifestation_of
Not even real nodes!
Strangely enough, a build from the same time last year (20230313) contains the same OMIM nodes and no more.
This is what MONDO:0026765 looks like in the most recent KG-Phenio:
MONDO:0026765 biolink:Disease congenital disorder of glycosylation, type IIr Graph CDG IIr|CDG2R|CONGENITAL DISORDER OF GLYCOSYLATION, TYPE IIr|congenital disorder of glycosylation, type IIr, X-linked recessive http://purl.obolibrary.org/obo/MONDO_0026765
If I look at a build from waaaay back in 2022 - xrefs are there!
MONDO:0026765 biolink:NamedThing congenital disorder of glycosylation, type IIr Graph CONGENITAL DISORDER OF GLYCOSYLATION, TYPE IIr|CDG2R|CDG IIr OMIM:301045 owl:Class
For comparison, this is what that node looks like in the most recent KG-OBO version of MONDO:
MONDO:0026765 biolink:Disease congenital disorder of glycosylation, type IIr OMIM:301045 mondo.json CONGENITAL DISORDER OF GLYCOSYLATION, TYPE IIr|CDG IIr|CDG2R|congenital disorder of glycosylation, type IIr, X-linked recessive http://purl.obolibrary.org/obo/MONDO_0026765 https://omim.org/entry/301045 rare|inferred_rare
So somewhere along the way, the xrefs (and also skos?) started being ignored.
I think the transform step in which node sources are added is not including xrefs. See https://github.com/Knowledge-Graph-Hub/kg-phenio/blob/master/kg_phenio/transform_utils/phenio/phenio_node_sources.yaml
OMIM xrefs (and others) are back in kg-phenio:
$ grep MONDO merged-kg_nodes.tsv | grep OMIM | wc -l
10359
In reference to https://github.com/monarch-initiative/monarch-app/issues/652 -
MONDO identifiers don't appear to have their corresponding OMIM IDs, or at least they don't end up in the Monarch KG. They appear to be populated in PHENIO as expected - see https://github.com/monarch-initiative/phenio/issues/67 Are they going missing in the KG-Phenio build?