Ontology meeting 2017-05-26

From GO Wiki
Jump to: navigation, search

Conference Line

Agenda

github Ontology Projects

Discussion points (Pascale):

  1.  NTR RO: 'transports' https://github.com/oborel/obo-relations/issues/161
  2.  Keeping old stanzas as synonyms for merged terms: Could this rule be relaxed? Sometimes the other name is not appropriate.
  3. Otherwise- is it possible to move secondary IDs ? For example "GO:0019060 intracellular transport of viral protein in host cell " (https://github.com/geneontology/go-ontology/issues/13478) has synonyms
    • viral capsid transport in host cell nucleus,
    • intracellular viral protein transport,
    • intracellular viral capsid transport and
    • viral capsid transport in host cell cytoplasm

Presumably the merge was done to the wrong term, these seem like synonyms for 'GO:0075733 intracellular transport of virus'

GO Ontology Editors Documentation

  • Moni has deployed the GO editors docs on RTD, available at http://go-ontology.readthedocs.io/en/latest/
    • It's not yet complete; e.g. we should add a page about how you edit the docs themselves. Now we have everything in place, it's easy!
    • All of the pages are an individual md page in our repo. See: https://github.com/geneontology/go-ontology/tree/master/docs E.g. https://github.com/geneontology/go-ontology/blob/master/docs/Installgit.md
    • These can be edited like any other file on github. You can do this from your local copies of the files, but you may find it more convenient to edit via the github web interface, just click the pencil icon. After the changes go into master, it gets deployed on RTD!
    • Adding new pages is also easy, do this the usual way, and make sure the TOC in index.rst is updated.
    • David OS - do you think it makes sense to fold in the auto-generated pattern docs here?

GOC Meeting

  • Review agenda items

Minutes

  • On call: Chris, David OS, Eric, Harold, Jim, Karen, Kimberly, Moni, Pascale, Suzi, Tanya

Adding 'transport' as a relation in RO

  • The existing relation, 'transports or maintains localization of', may be more than is needed for some terms
  • A simpler 'transport' relation could be added as a child of 'transports of maintains localization of'
    • This would not, presumably, break things
    • We do still need the 'maintains' concept in a parent relation, though
  • There is also import and export and these are not used quite consistently throughout the ontology
    • The high level terms for import and export do not specify from where to where, so are not very useful.
    • Their usage should be reviewed.
    • Specific import and export terms can be defined entirely wrt start and end points - no need for specific relations.
  • How best do we distinguish maintenance vs transport.
    • The definitions for each branch are now reasonably clear. But guidance for evidence may be tricky.
  • This is an area of GO that may be overly complicated
    • Localization is mixed with location
    • It will be a pretty big job to untangle all of this.


  • AI: Make new 'transports' and 'occurs across' relations. The latter will replace 'results in transport across'

https://github.com/oborel/obo-relations/issues/166

Old term names as synonyms

  • Right now, there is a check that requires the merged term name to be kept as a synonym
  • But, sometimes, the new term name might not be appropriate
  • Or, if two terms were incorrectly merged, then you might want to remove some synonyms
  • Some term names are actually confusing, and if the term is merged, then it might be nice to have the option to remove the term name altogether
  • Is it worth keeping the check in just in case people forget to add them?
  • Yes, since there's another way around removing synonyms

Viral processes

  • Pascale has worked on this branch of the ontology to make sure that viral proteins are not annotated to cellular processes
  • Kimberly and Pascale will check on some existing annotations

Design patterns for compound functions

  • Using has_part relations for receptor terms fails to classify receptors with receptor and effector functions and would break GO-CAM inferences
  • Solution is to use has_component relations and its children
  • Using ATPase as an example, the annotated entity would be inferred to be an ATPase, but that relation would not exist in the ontology
  • Using transcription factor example, has_necessary_component DNA binding
    • Side discussion about representing protein complexes in Noctua - what will best practices be?
    • Side discussion about Noctua template for transcription
    • In order to get the proper reasoning, brought back a superclass of 'transcriptional regulator' activity

Editors Documentation

  • Now on Read the Docs
  • Please double-check and send any feedback to Moni
  • Still some comments on the Google doc that are unresolved
  • At this point, edits should be made on the md files on github and then generate a pull request