6 OMIM ids result in an "Internal Server Error" when submitted to the similarity endpoint

[anas-elghafari]

Each of the following 6 OMIMs result in an internal server error (response status 500) when submitted as a param to the similarity endpoint:
OMIM:256700 OMIM:175100 OMIM:300494 OMIM:175505 OMIM:259660 OMIM:138800

Example to reproduce the server error:
curl -X 'POST'
'https://openpredict.semanticscience.org/similarity'
-H 'accept: application/json'
-H 'Content-Type: application/json'
-d '{
"subjects": ["OMIM:138800"],
"objects": [],
"options": {}
}'

Additionally, while all of these show up in the similarity hits of other OMIM ids, they -alone among the 308 results- never have an 'object_label'.

[dagshub]

Join the discussion on DagsHub!

[micheldumontier]

probably the way the service works is that drugs go into the subject slot and diseases go into the objects slot. however, when i try putting the OMIM ids in the objects slot, there are no results.

[anas-elghafari]

I tried a couple of other examples, and there was no problem witth placing omim codes in the "subjects" slot (example below). It is only with those 6 that an internal server error was triggered

curl -X 'POST'
'https://openpredict.transltr.io/similarity'
-H 'accept: application/json'
-H 'Content-Type: application/json'
-d '{
"subjects": ["OMIM:246300"],
"objects": [],
"options": {}
}'

Will try to debug on the server

[anas-elghafari]

Update: tracing this issue, it is due to the trapi_predict_kit package which is not failing gracefully when the RENCI API cannot resolve an OMIM (RENCI API returns a "null" value but trapi_predict_kit looks for a list, triggering an TypeError which cascaded through this package).

On the code side, the trapi_predict_kit should be fixed to handle failures better.

On the data integrity/data quality issue: RENCI's data is missing between 50-100 OMIMs (estimate based on sampling. Exact number requires testing every entry). Compare the RENCI response to two OMIMs (246300, 256700):

https://nodenormalization-sri.renci.org/1.5/get_normalized_nodes?curie=OMIM%3A246300 [found]

https://nodenormalization-sri.renci.org/1.5/get_normalized_nodes?curie=OMIM%3A256700 [missing from the data]