Open sstemann opened 2 years ago
@sstemann Apologies for the unassigned and re-assign; I was trying to assign @finnagin, but adding an assignee apparently unassigned everyone
Isn't biolink:ChemicalSubstance
deprecated now, or are we still supposed to be supporting old biolink versions? COHD is responding
biolink:ChemicalSubstance was not recognized as a biolink category
For some reason, OpenPredict is responding to the gene node with diseases.
@CaseyTa This question has come up with ARAGORN/COHD integration, too. "Chemical substance" is officially an alias of "small molecule" in the biolink model: https://github.com/biolink/biolink-model/blob/2.1.0/biolink-model.yaml#L7250. Does that mean that we should continue to accept it and treat it as synonymous with "small molecule"? Or do we ignore those aliases?
Regardless, it would be convenient it servers could simply ignore categories that they do not recognize/support, rather than throwing an error. For example, it would be nice if nodes with "categories": ["biolink:SmallMolecule", "biolink:ChemicalSubstance"]
did not cause the server to abort.
@patrickkwang, we're validating the input categories using the Biolink Model Toolkit, so bmt is telling us that biolink:ChemicalSubstance
is not a category.
Regardless, it would be convenient it servers could simply ignore categories that they do not recognize/support, rather than throwing an error. For example, it would be nice if nodes with
"categories": ["biolink:SmallMolecule", "biolink:ChemicalSubstance"]
did not cause the server to abort.
If this issue is coming up often enough, we can make that change in COHD. Does this arise because we're trying to run a lot of old queries that were previously created under older Biolink versions? Or is this coming from different services running different versions of Biolink which support different categories?
BTE returns the expected results with either "categories": ["biolink:SmallMolecule"]
or "categories": ["biolink:SmallMolecule", "biolink:ChemicalSubstance"]
. We in principle could follow the alias in biolink model, but I think we have higher priorities before the Sept relay and probably the Dec demo too. So I'd prefer if our queries strictly used the v2.2.x classes...
@sierra-moxon is this something BMT should help with? ^^
My mistake, I meant to change this after the last standup. I've resubmitted with SmallMolecule bd5bba27-3d28-425f-a05d-6ff802833c4a
Same issue with the SmallMolecule vs ChemicalSubstance for us
Edit: thanks Mark, that ones works for us.
The error has been fixed in ARAX (thanks @finnagin )
Query: workflow2.json PK: bd5bba27-3d28-425f-a05d-6ff802833c4a (deprecated a0b7182c-d4fc-4d47-87b0-d84ac3861e1c) Control:
Results Tracking Sheet