we'd be pretty safe in general giving the label associated with the lowest EG ID
In many cases, there is one gene name associated from SwissProt (you are using this as both "label" and "name" right now in your SP namespace). We should canonicalize/decanonicalize to the HGNC/EGID that matches that one, which generally seems to follow rule #1 as well.
read through. We want to use the non-readthrough gene/protein label. These SP entries are only associated with one gene name and one HGNC.
two essentially identical proteins, coded for by different genes (IFNA1/IFNA13, several defensins). It would probably be easier to use the 1st identified gene as the label, since this is what will be in the lit.
not sure what's going on (ANXA8). There are two gene IDs, but one of them (for ANXA8L1) also has its own SP entry. Safe to treat similar to the read-throughs, but may have more untangling to do.
Add SFAM nanopub for hasMembers: https://github.com/OpenBEL/openbel-framework-resources/blob/latest/resource/protein-families.bel
Equivalence to MESH/CHEBI and migrate from SCHEM https://github.com/OpenBEL/resource-generator/blob/master/datasets/selventa-legacy-chemical-names.txt
Equivalence to GO complexes and migrate from SCOMP https://github.com/OpenBEL/resource-generator/blob/master/datasets/selventa-named-complexes.txt
Update Swissprot to handle multiple gene parents