Ontology meeting 2016-04-28

From GO Wiki
Jump to navigation Jump to search

Attendees: Melanie, Paola, Harold, Heiko, Stan (first part of call), Tanya, David H, Chris, David OS

Minutes: David OS


Follow up: More 'response to non-physiological chemical stimulus' terms coming in via TG

Some recent requests: response/cellular response to ouabain and response/cellular response to Dizocilpine

https://en.wikipedia.org/wiki/Ouabain

https://en.wikipedia.org/wiki/Dizocilpine

We rejected those term requests on the same grounds as previously. Chris suggested we invite Stan to attend an editors' call for discussion and to look for viable alternatives to RGD needs. Stan will join us today.


 Stan: We run a semi-automated pipeline to mine gene:chemical interactions from CTD.
 This provides biological process annotations - often for genes that have no other annotation.
 Chris: what type of interactions are you getting from CTD?  Binding interactions, changing expression?
 And does CTD distinguish these?

 Stan: It's a mix of these and others.  They do make some distinctions between different types of interaction.
 Q: What's your use case?  Is it sufficient for you to display the CTD associations directly?
 Stan: We can do that instead of submitting non-physiological response terms in future.  
 Paola: If you have cases where there may be a valid annotation to GO terms from these data, 
 you can submit a ticket rather than using TG-FF.  This will allow us to discuss how best to meet
 your needs

Protein Complex WG

Pilot project to remove large numbers of leaf nodes & infer annotation cancelled.

https://github.com/geneontology/protein-complex-refactor/issues/17

 Discussion:  There is still no clear criteria for how many species a complex has to be in for it to stay in the ontology.
 Reasons to state that two complexes in different species are 'the same' are likely to vary, so no clear rules for this either.
 For these reasons, it will be hard for editors to review and judge what should stay without outside help.
 DOS: Where we know all the gene products that make up a complex in species X 
- because this is specified by IntAct - shouldn't we limit direct annotation to those gene products, 
 with any additional annotations using co-located_with qualifier.
 AI: Write guidance on submission of new terms:  
   We will not accept terms unless the submitter gives examples in multiple species. 
 AI: Editors will not attempt to assess species generality of existing terms, but
  will take feedback from protein complex WG on which complexes are known to be multi-species.
   - We will make sure synonyms (with appropriate refs) are in place to cover names in other species
   - Modify defs/comments to clarify basis for grouping across species 
      &/or enumerate examples from multiple species
 - Removing terms that are considered to be too species specific, as advised.
 AI: DOS to follow up on WG meeting plans with Brigit.

Inheriting taxon rules from UBERON

See https://github.com/geneontology/go-ontology/issues/12411#issuecomment-212806374

 AI: Chris to turn on uberon taxon constraint import.  If too many
 issues to fix, we will turn import off and use the OWL inconsistent class module file
 produced by Jenkins to guide fixes until it is safe to turn the import back on.

Switching to new inference cycle with imported inference file

Updated checklist