This could improve mapping/merging entity references (e.g. in IntAct, DIP, etc. PSI-MI data converted to BioPAX) and thus get more HGNC Symbol and UniProt IDs available for analysis and converting to GSEA, SIF formats.
But we have to be careful with this, because there are cases, such as ND5 (Recon X data), when a single HGNC Symbol maps to hundreds (in fact >1000) UniProt IDs (this "kills" our SIF rules/converter)...
(There are also original protein references (or PSI-MI interactors), which lonk to different canonical proteins, sometimes different species, which should be modelled either as generic PR or complex, but it's hard to tell/infer for sure in a Converter/Cleaner/Normalizer, based on any standard term, property, etc...)
IgorRodchenkov
commented
5 years ago
This could improve mapping/merging entity references (e.g. in IntAct, DIP, etc. PSI-MI data converted to BioPAX) and thus get more HGNC Symbol and UniProt IDs available for analysis and converting to GSEA, SIF formats. But we have to be careful with this, because there are cases, such as ND5 (Recon X data), when a single HGNC Symbol maps to hundreds (in fact >1000) UniProt IDs (this "kills" our SIF rules/converter)... (There are also original protein references (or PSI-MI interactors), which lonk to different canonical proteins, sometimes different species, which should be modelled either as generic PR or complex, but it's hard to tell/infer for sure in a Converter/Cleaner/Normalizer, based on any standard term, property, etc...)