edeutsch
commented
4 years ago okay, I confirm, that I'm not seeing this in master.
I can repro in ranker-refine?
Closed dkoslicki closed 4 months ago
edeutsch
commented
4 years ago okay, I confirm, that I'm not seeing this in master.
I can repro in ranker-refine?
edeutsch
commented
4 years ago I switched my branch to ranker-refine and am seeing a bazillion of these errors:
2020-09-09 16:53:17.522553 DEBUG: Querying Columbia Open Health data for info about neutropenia and Hypoplastic fingernail
2020-09-09 16:53:17.522729 ERROR: Traceback (most recent call last): File "/mnt/data/orangeboard/devED/RTX/code/UI/OpenAPI/python-flask-server/swagger_server/../../../../ARAX/ARAXQuery/Overlay/overlay_clinical_info.py", line 119, in make_edge_attribute_from_curies res = self.cohdIndex.get_concept_ids(source_curie) File "/mnt/data/orangeboard/devED/RTX/code/ARAX/ARAXQuery/Overlay/../../KnowledgeSources/COHD_local/scripts/COHDIndex.py", line 225, in get_concept_ids cursor.execute(f"select distinct t1.curie, t1.concept_id from CURIE_TO_OMOP_MAPPING t1 inner join CONCEPTS t2 on t1.concept_id = t2.concept_id where curie='{curie}';") sqlite3.OperationalError: no such table: CURIE_TO_OMOP_MAPPING
2020-09-09 16:53:17.522744 ERROR: Something went wrong when adding the edge attribute from COHD.Any ideas on how to address that?
edeutsch
commented
4 years ago @dkoslicki I am not able to test further because of previous post's problem.
BUT, I am going to guess that the error arises because you have a confidence value that cannot be converted to a float. I would check all the confidence values in your resulting message, and I'll wager that one of them is not convertible to float. If you fix that, it will likely work?
I can make the code more robust to not fail so poorly when that happens, but it shouldn't happen I suppose.
dkoslicki
commented
4 years ago @edeutsch Right: the errors you see with COHD are due to updates in the COHD local database, so that gets fixed if you merge and then run the code. MWE:
git checkout ranker-refine
git pull origin ranker-refine
git merge --no-ff origin/master
<fix the merge conflict>
git add code/ARAX/ARAXQuery/Overlay/predictor/predictor.py
python ARAX_query.py 16Let me know if that works in reproducing the error for you
edeutsch
commented
4 years ago @dkoslicki I ran aground trying the above git commands. I could not get it to work.
BUT, I checked in a fix to master than now checks to see if confidence scores can be converted to floats in a robust way. If any confidence score cannot be converted to a float, instead of erroring out (like it was), it prints this message to STDERR:
eprint(f"WARNING: Confidence value '{result.confidence}' cannot be converted to float")
and then result.confidence is set to -999.0
So be on the lookout for confidence scores of -999.0 and consider that something that needs to be fixed in the ranker. All output confidences should be floats.
edeutsch
commented
4 months ago closing ancient history.
@edeutsch I discovered this issue when merging
masterintoranker-refine(one conflict, a single comment, easily fixed) and when running the DSL:And trying to run the DSL in a local UI, I get the error pasted below (which I didn't observe on
master). Note that pretty much anything that usesreturn(message=true, store=true)throws this error...I find this especially odd since
RTXFeedback.pywasn't touched in the merge:This issue blocks #941