Closed cbizon closed 3 years ago
It looks like these are coming from biocarta / reactome / etc but proxied through ConsensusPathDB and mygene.info. @colleenXu it looks like you added the relevant lines to https://github.com/NCATS-Tangerine/translator-api-registry/blame/master/mygene.info/openapi_full.yml -- do you have any insights here to share?
Example edge:
{
"predicate": "biolink:participates_in",
"subject": "NCBIGene:1017",
"object": "name:e2f1 destruction pathway",
"attributes": [
{
"name": "provided_by",
"value": [
"ConsensusPathDB"
],
"type": "biolink:provided_by"
},
{
"name": "api",
"value": [
"MyGene.info API"
],
"type": "bts:api"
},
{
"name": "publications",
"value": [],
"type": "biolink:publication"
}
]
}
@andrewsu I believe these are BIOCARTA pathway names. This is likely because the BIOCARTA ID is not included in the id ranking for Pathway entities here.
Including it in the id ranking before "name" will likely fix this issue.
@marcodarko do you have in your dev environment an easy way to test whether @colleenXu's pull request fixes the issue? Specifically, in response to the TRAPI query in the first comment, we should see no nodes with name:
as the prefix.
@andrewsu yes that seems to solve this issue, I'll make a PR to our code looks like this was made to Colleen's forked code.
For this query (AS edited 2021-06-20 -- updating to TRAPI v1.1):
We get back a bunch of results, most of which are KEGG or REACT or WIKIPATHWAY, but we also get back these, which look like an error, or at least are not biolink compliant ids: