Ontology meeting 2020-02-03: Difference between revisions

From GO Wiki
Jump to navigation Jump to search
Line 56: Line 56:
*Do we still want molecular function regulator as an MF class in the ontology?  Should we keep its logical definition?
*Do we still want molecular function regulator as an MF class in the ontology?  Should we keep its logical definition?
*Do we want to export inferred molecular function regulator annotations, even for upstreams gps who activity directly regulates that of downstream gps?
*Do we want to export inferred molecular function regulator annotations, even for upstreams gps who activity directly regulates that of downstream gps?
====Inferred Protein Binding Annotations====
*We still get inferred protein binding annotations when connecting two MFs with a directly positively regulates relation and the second MF is enabled by a protein entity


==Other Business==
==Other Business==

Revision as of 16:37, 30 January 2020

Conference Line

See calendar for details

On call:

Agenda

Discussion topics

Aligning Ontology, Annotation, ShEx

  • We are creating ontology term-specific shapes in the ShEx schema
  • The goal is to align what is used for annotation with the equivalent expression in logical definitions of the ontology
  • We are also moving towards only using RO relations in GO, i.e. retiring, or incorporating into RO, GOREL relations or RO-pending relations

GOREL relations proposed to RO

GOREL relations that could potentially be retired and replaced with existing RO relations

  • ACTION: As a pilot, David and Kimberly replaced 'occurs in' in cell division logical definitions with 'has input' for specific cell types
  • https://github.com/geneontology/go-ontology/pull/18690
  • Also removed some asserted parents for cell division terms - this changed some parentage wrt neuroblasts and somatic stem cell parentage (but is now consistent with CL)
  • Overall, this wasn't too difficult and we could continue along these lines for the other relations in question (results in fission of, results in fusion of, results in maintenance of, results in distribution of)

From 2020-01-27 call:

'results in division of' (GOREL:0001019)
  • Not used in the ontology
    • In the ontology, logical definitions use:
      • 'occurs in', e.g. cystoblast division = cell division and 'occurs in' some cystoblast
      • 'has input', e.g. germ-line stem cell division = cell division and 'has input' some germ-line stem cell
      • loosely, but not exclusively, along somatic vs stem cell lines
    • No design pattern (that I could find)
  • For annotation, it looks like curators use 'occurs in', 'has participant, 'results in division of' although there aren't many annotation extensions at all for 'cell division' and children
  • PROPOSED ACTION: finish review of ontology diff
  • PROPOSED ACTION: obsolete 'results in division of' and 'regulates o results in division of'
  • PROPOSED ACTION: create a design pattern for cell division using 'has input'

Follow-up from annotation call

Binding terms

  • Curators asked for firm guidelines on when to request new 'binding' terms
  • Can we get back to them with a proposal? For example:
    • Going forward, curators will annotate binding using an appropriately general genus term (TBD) and the 'has input' extension for the chemical entity
    • We can generate monthly reports of annotation extension use to see if there are any highly (TBD) used extensions for which we should create ontology terms
    • Things to consider:
      • For chemical entities, what ontology terms should curators use?
      • What should we do with existing terms in the ontology?
      • We still need to convey what binding events we want to capture as MFs, and why, in GO

Compound functions and molecular function regulators

  • We need to decide on a design pattern for compound functions
  • For the estrogen nuclear receptor, the demo GO-CAM model had estrogen binding 'directly positively regulates' nuclear receptor activity
  • This results in an inferred molecular function regulator annotation on the estrogen binding node meaning that the estrogen receptor is a molecular function regulator of itself, which curators felt was not biologically correct/intuitive and they wouldn't want to show this to users
  • Can we model compound functions with 'has part' in the ontology rather than a regulatory relation between the separate functions?
  • Note that we will still get many inferred molecular function regulator MF annotations for models where we use the 'directly positively/negatively regulates' relation between two activities 'enabled by' different gene products
  • Do we still want molecular function regulator as an MF class in the ontology? Should we keep its logical definition?
  • Do we want to export inferred molecular function regulator annotations, even for upstreams gps who activity directly regulates that of downstream gps?

Inferred Protein Binding Annotations

  • We still get inferred protein binding annotations when connecting two MFs with a directly positively regulates relation and the second MF is enabled by a protein entity

Other Business

https://github.com/geneontology/go-ontology/projects/1

Minutes should be recorded on individual tickets and in individual projects