biolink-api
biolink-api copied to clipboard
"category" field in /nlp/annotate/entities is always empty
When querying against /nlp/annotate/entities, the returned "category" field for each token is always an empty list.
For example,
curl 'https://api.monarchinitiative.org/api/nlp/annotate/entities' \
--data-raw 'content=Lewis+%281978%29+found+7+affected+males.'
results in (partial result):
{
"content": "Lewis (1978) found 7 affected males.",
"spans": [
...
{
"start": 30,
"end": 36,
"text": "males.",
"token": [
{
"id": "UBERON:0003101",
"category": [],
"terms": [
"male organism"
]
}
]
}
...
]
}
Querying directly against Scigraph Ontology's API, e.g. via:
curl 'https://scigraph-ontology.monarchinitiative.org/scigraph/annotations/entities' \
--data-raw 'content=Lewis%20(1978)%20found%207%20affected%20males.'
produces (again, in part):
[
...
{
"token": {
"id": "UBERON:0003101",
"categories": [
"anatomical entity"
],
"terms": [
"male organism"
]
},
"start": 30,
"end": 36
}
...
]
I'm not sure, but it seems likely that this is because the field in Token (https://github.com/biolink/ontobio/blob/31658ad37eb02f0e061ca5c4de36618cab6ea02b/ontobio/model/nlp.py#L14) is named category, not categories as SciGraph's API is returning.
To fix this, we'd need to modify the category field name in ontobio.model.nlp.Token to be categories, update biolink's ontobio dependency to this new version, then redeploy biolink.
In addition to renaming category to categories in ontobio.model.nlp.Token, the matching serializer class in biolink also needs to be updated in the same way: https://github.com/biolink/biolink-api/blob/master/biolink/datamodel/serializers.py#L336.
To add clarity why this is needed for the frontend:

It causes the annotations here to have (?) icons, because I dont know the category of each annotation, because that field is blank.