Closed saramsey closed 1 year ago
Confirmed, this same problem is showing up in KG2.8.3c (confirmed on kg2canonicalized.rtx.ai
).
I wonder if we just need to add an if pmid != ''
block here:
Another problematic edge (reported by Kaiwen He from Team Unsecret Agent):
NCBIGene:79092---biolink:gene_associated_with_condition---None---None---None---UMLS:C2930842---DisGeNET: False UMLS:C2930842 biolink:gene_associated_with_condition gene_associated_with_condition infores:disgenet PMID: {} gene_associated_with_condition biolink:gene_associated_with_condition NCBIGene:79092 2018 biolink:gene_associated_with_condition NCBIGene:79092 UMLS:C2930842
It looks like this issue is fixed in KG2.8.4pre
, so I am going to close out this issue:
match (n)-[e]-(m) where e.id="NCBIGene:79092---biolink:gene_associated_with_condition---None---None---None---UMLS:C2930842---DisGeNET:" return n, e, m limit 1
{
"predicate": "biolink:gene_associated_with_condition",
"domain_range_exclusion": "False",
"negated": "False",
"primary_knowledge_source": "infores:disgenet",
"relation_label": "gene_associated_with_condition",
"publications_info": "{}",
"subject": "NCBIGene:79092",
"source_predicate": "biolink:gene_associated_with_condition",
"predicate_label": "gene_associated_with_condition",
"id": "NCBIGene:79092---biolink:gene_associated_with_condition---None---None---None---UMLS:C2930842---DisGeNET:",
"update_date": "2018",
"object": "UMLS:C2930842"
}
In RTX-KG2.8.0.1c, I am seeing the following edge, which I think is not valid due to the "empty" CURIE for the PMID. The following Cypher query (on
kg2canonicalized2.rtx.ai
):produces:
I don't think the
PMID:
is permitted. This issue was first brought to my attention by the Translator Feedback group (https://github.com/NCATSTranslator/Feedback/issues/414)