In the last run, we ran into a case where a PUBCHEM identifier was mentioned in RxCUI but not included in our PubChem download. To account for such cases, I've modified the DrugChemical conflation code so that if either the subject or the object cannot be normalized, we skip this pair entirely. This allows us the pipeline to complete while we figure out what happened with that PubChem identifier.
Also, we previous threw an exception if the subject identifier isn't mapped to an RxCUI, but now we log this as a warning and skip the mapping.
In the last run, we ran into a case where a PUBCHEM identifier was mentioned in RxCUI but not included in our PubChem download. To account for such cases, I've modified the DrugChemical conflation code so that if either the subject or the object cannot be normalized, we skip this pair entirely. This allows us the pipeline to complete while we figure out what happened with that PubChem identifier.
Also, we previous threw an exception if the subject identifier isn't mapped to an RxCUI, but now we log this as a warning and skip the mapping.