Open Genomewide opened 1 year ago
Thanks for reporting this issue!
The development NodeNorm is out of sync with the production NameRes, since we're testing a new release, but the production NodeNorm (https://nodenorm.transltr.io/1.3/get_normalized_nodes?curie=LOINC%3AMTHU015779&conflate=true) doesn't have LOINC:MTHU015779
. However, I'm not sure where LOINC:MTHU015779
is coming from -- if you search the production NameRes for interleukin 2
, you get a bunch of results including PUBCHEM.COMPOUND:130168 and NCBIGene:3558 (https://name-lookup.transltr.io/lookup?string=interleukin%202&offset=0&limit=10). Both of those identifiers refer to the small molecule and gene form of interleukin-2 respectively (https://nodenorm.transltr.io/1.3/get_normalized_nodes?curie=PUBCHEM.COMPOUND%3A130168&curie=NCBIGene%3A3558&conflate=false).
So, I think the big questions is: where is ARAX's synonym search getting LOINC:MTHU015779, and how can we stop it? I had a look at the RTX repository, and it looks to me like it generates synonyms by itself. I've opened an issue with ARAX (https://github.com/RTXteam/RTX/issues/1920) and will follow up once I hear back from them.
I think I found a normalization bug. Node normalizer returns nothing for either of these curies (FMA:84051 or LOINC:MTHU015779) for interleukin-2:
From the smart API:
They are also disconnected on the ARAX synonyms search (if that is run on something different).
For FMA:84051 biolink:Protein Interleukin-2 is a named thing. This is different on the arax 'synonyms' than if you search for 'interleukin 2' which returns biolink:NamedThing and LOINC:MTHU015779?
So, the terms are unconnected and should be. An d you can only get to the ARAX FMA information is by searching with the FMA CURIE.