Incorrect annotations to CHAC1 and CHAC2 from Reactome annotation

[cmungall]

CHAC1 and CHAC2 have confusing seemingly conflicting stories when Reactome annotations are combined from elsewhere:

https://www.uniprot.org/uniprotkb/Q9BUX1/entry

MF:

BP:

The MF one is incorrect and comes from an incorrect Reactome mapping: https://reactome.org/content/detail/R-HSA-6785928

It should be mapped to

• Glutathione-specific gamma-glutamylcyclotransferases (consistent with the FlyBase-provided annotation)

Not to

• gamma-glutamylcyclotransferases

Two asides here:

• shouldn't we be catching these as part of the global RHEA-EC-GO-Reactome reconciliation? cc @balhoff
• The names make it seem like these two MFs should be related via is-a, but in fact I think the current sibling relationship is correct. EC terminology is odd...

ACTIONS:

-[ ] Fix the Reactome MF mapping

The above is a straighforward incorrect mapping. the next is more nuanced.

CHAC1/2 is involved after synthesis of GSH, when GSC is recycled. According to GO BP start/end rules, anything happening after synthesis should not be annotated to synthesis. The current GO Central annotation (catabolism) is correct. the conflicting Reactome-derived annotation is IMO incorrect. Where does it come from?

If we walk up the Reactome tree:

We get to Glutathione synthesis and recycling
https://reactome.org/content/detail/R-HSA-174403

And this is mapped to the GO BP:

I think as a loose mapping this is fine. The processes are clearly related. But it's not a propagatable (exact or narrow-to-broad) mapping, and treating it as such gives confusing annotations.

But in MetaCyc terms, synthesis+recycling would be a superpathway of syn+catab, but we don't have such a term in GO (in general we don't have superpathway/composite/mereological sum terms).

ACTIONS:

• Fix the Reactome BP mapping. Options:
  • map to a broader GO term (GSH metabolism is better but it's not actually the correct semantics, since this is a superclass not superpwathway)
  • remove the mapping
  • map the individual steps to BPs (I think this is best)
  • use a non-exact mapping predicate (would likely require changes in the Reactome database)

@deustp01, I assigned you since I believe these need to be activated at the Reactome end, but lmk if I should assign someone else.

This touches on a lot of things that have been discussed on the ontology calls lately with @pgaudet @sjm41, but I put in the annotation tracker.

[cmungall]

For my own notes:

This was spotted using an automated QC procedure that aligns GO BP, MF annotations, RHEA mappings, RHEA to CHEBI, and GO BP logical definitions: