search

geneontology / pathways2GO

Code for converting between BioPAX pathways and Gene Ontology Causal Activity Models (GO-CAM)
8 stars 0 forks source link

Ensure used EC2GO mapping is to GO function term #140

Closed dustine32 closed 1 year ago

dustine32 commented 2 years ago

Use case: SGD pathway - [formaldehyde oxidation II (glutathione-dependent)](https://pathway.yeastgenome.org/YEAST/NEW-IMAGE?type=NIL&object=PWY-1801-1&redirect=T) Reaction 1.2.1.2-RXN has EC number 1.17.1.9, which is mapped to both formate dehydrogenase (NAD+) activity (GO:0008863) and formate dehydrogenase complex (GO:0009326). Of these two GO terms, only GO:0008863 is a function. The pathway conversion code should filter out any non-function terms to prevent invalid OWL situations like this: image Where a formate dehydrogenase complex -part_of-> biological process.

@ukemi @deustp01 We probably discussed this already, but can you clarify whether or not EC's should ever be mapped to non-molecular function GO terms? If they are supposed to be MF-only, I will place this code change where it will be more widely used.

deustp01 commented 2 years ago

@ukemi I'm clueless on ythis EC question - I'd tag Harold but forget his tag.

dustine32 commented 2 years ago

Thanks @deustp01! Paging @hdrabkin on this!

ukemi commented 2 years ago

@dustine32 it depends (doesn't it always?). When this file was first made over 20 years ago, the goal was to simply generate GO annotations for genes that had EC cross references at resources like MGI. With that goal in mind, some EC identifiers could be mapped to CC and BP terms if the enzyme described in the EC entry was enabled by a gene product and we could safely say that the gene product was part of a complex that carried out that MF or was always a part of a BP. The file has probably evolved a lot since we first made it, but that was the original intent. That said, for your purposes I would suggest that you only take EC->MF information for the translation. At the end of the day, since we are now including GP to Term relationships in all annotations we should rethink how these translation files are constructed. I think it would be prudent to include those relations if the files are still of value. @vanaukenk don't you think?

dustine32 commented 2 years ago

Ah, OK. I will just be careful and heavily comment that I'm filtering for EC-mapped MF terms only. No problem!

Thanks for the response @ukemi!

dustine32 commented 1 year ago

Closing since the fix has been merged for a while and running in the latest batch of YeastPathways models.