Annotation Conf. Call 2020-02-04

From GO Wiki
Jump to: navigation, search

Agenda and Minutes

  • Present: David, Dmitry, Edith, Giulia, Harold, Helen, Karen, Kimberly, Li, Midori, Patrick M, Rob, Ruth, Sabrina, Seth, Stacia, Suzi A, Tanya, Tremayne

Annotation Calls

  • We will meet weekly on Tuesdays at 8am PST.
  • Will no longer make the distinction between annotation and GO-CAM/Noctua calls.

Noctua

  • Noctua is now updated to make the new Noctua form the default version of the form.
  • Please let us know if you experience any trouble with Noctua after the update, or have questions on how to use the form.
  • Documentation is here: http://wiki.geneontology.org/index.php/Noctua
  • Thank you to the MGI curators and to Sabrina for their feedback

Annotation Relations

  • Goal is to review use of relations across annotations (standard and GO-CAM) and the ontology to ensure consistency wherever possible
  • Relation discussed this week:

Follow-Up Discussion with Ontology Editors

Chemical Entity Binding Terms

  • Question from last week's call: should curators request new pre-composed binding terms for binding chemical entities, or should they use 'binding' has_input 'ChEBI term'?
  • Ontology editors discussed this on their weekly call yesterday and the decision is:
    • Going forward, curators are asked to annotate to 'binding' and use the 'has input' extension to capture the chemical entity
    • Future work will evaluate existing 'binding' terms in the ontology and decide which should stay and also the criteria for instantiating new 'binding' terms
    • As an annotation group, we will also continue to evaluate what 'binding' annotations are appropriate for GO and why
    • Most groups can make annotation extensions in their current tools; if not, Noctua is available for groups to use partially or fully
Discussion
  • Using post-composed terms for chemical binding raised several issues, mostly around the potential difficulties in using a large ontology such as ChEBI and the effects that would have on curators (GO and community) trying to find the correct term for annotation.
  • One possible solution is to provide a snapshot of ChEBI for use in GO annotation tools.
  • It would be good to review how ChEBI is currently used in GO: https://www.ncbi.nlm.nih.gov/pubmed/23895341
  • Inherent in this discussion is the question of what binding annotations should be made and why. This still needs to be clarified.

Molecular Function Regulators

  • One of the inferred annotations generated from the proposed nuclear receptor model was that of molecular function regulator (GO:0098772)
  • Molecular function regulator
    • Textual definition: A molecular function that modulates the activity of a gene product or complex. Examples include enzyme regulators and channel regulators.
    • Logical definition: molecular function AND ('directly regulates' some molecular function)
    • Any upstream MF that directly regulates a downstream MF will be inferred to be a 'molecular function regulator' by the reasoner
    • Two issues from last week's call:
      • Was the model of nuclear receptor activity correct?
        • The idea was to try to model cases where curators used the 'activated by' extension relation, but in the case of nuclear receptors the premise that the ligand binding 'activates' the transcription factor activity may not actually be the correct way to think about this compound function
        • Ontology editors are currently reviewing how compound functions like 'nuclear receptor' are logically defined
        • Also reviewing 'molecular function regulator' as a class in the ontology - is it still useful?
      • 'directly' in 'directly regulates' = direct physical interaction
        • reasoning rule behind-the-scenes that generates the resulting 'protein binding' annotation based on this meaning of 'directly'
        • keeping the rule for now, but may reevaluate in the future
    • Note that we have the ability to refine what is export in the GPAD file, even if those annotations are created via reasoning
      • This is an option for retaining the logical definitions in the ontology and the intended meaning of the relations without necessarily outputting 'standard' annotations that groups may not want to display
Discussion
  • The 'directly regulates' relations are only intended to be used between activities 'enabled by' different gene products.
  • This will prevent molecular function regulator annotations to gene products based on internal binding of small molecules.
  • There are still questions about annotating compound functions and where and how curators should place supporting evidence for the sub-functions.
  • We need to look at examples from papers to make sure we're all on the same page here.

Other Annotation Issues

  • 'has direct input' -> 'has input' (Helen emailed Alex, cc'ed Kimberly, Pascale)
  • use of other relations in annotation tools like Protein2GO need to be reviewed - if groups are using gorel as the source, we need to make sure relations no longer in use are obsoleted in gorel
  • 'occurs at' -> 'occurs in'
  • if curators using Protein2GO need bulk updates to relations such as these, they should contact Alex to either change the relation, or remove the extension