It looks like we can save ourselves a lot of mapping trouble by switching from Ensembl2Reactome.txt to NCBI2Reactome.txt, and we should also filter the species list down to match our species list.
Blocked awaiting a fresh Biolink Model release with PR's 1031 and 1033, which revised the model to accommodate elements which I used in the code PR for this issue.
It looks like we can save ourselves a lot of mapping trouble by switching from Ensembl2Reactome.txt to NCBI2Reactome.txt, and we should also filter the species list down to match our species list.
We'll need to change out the url for https://reactome.org/download/current/NCBI2Reactome.txt and change the prefix added to
NCBIGene
There are also a few non-integer entries in the component column that we either need to filter out or handle differently.
The alternative to this would involve things like making mapping files from yeast systematic names to SGD IDs.