Annotation Conf. Call 2020-04-07: Difference between revisions

From GO Wiki
Jump to navigation Jump to search
Line 24: Line 24:


=== Proposed GAF Update, GAF 2.2 ===
=== Proposed GAF Update, GAF 2.2 ===
* GAFs will continue to be produced and supported by the GOC into the foreseeable future
* GAF 2.1 will continue to be produced and supported by the GOC into the foreseeable future
* We are proposing an incremental update to the GAF, though, to allow for use of the full set of gp2term relations
* We are proposing an incremental update to the GAF, though, to allow for use of the full set of gp2term relations
* Proposed set of allowed relations is in this github ticket: https://github.com/geneontology/go-annotation/issues/2917
* Proposed set of allowed relations is in this github ticket: https://github.com/geneontology/go-annotation/issues/2917
Line 34: Line 34:
*Why do this?
*Why do this?
** The expanded set of gp2term relations is available in annotation tools, e.g. Noctua and Protein2GO, but by not including them in the GAF, we don't give the GOC, or users, any mechanism to filter specific sets of annotations for GAFs and subsequent analyses.  Making this change will allow GO and groups to do this.
** The expanded set of gp2term relations is available in annotation tools, e.g. Noctua and Protein2GO, but by not including them in the GAF, we don't give the GOC, or users, any mechanism to filter specific sets of annotations for GAFs and subsequent analyses.  Making this change will allow GO and groups to do this.


=== GPAD/GPI 2.0 Specifications ===
=== GPAD/GPI 2.0 Specifications ===

Revision as of 10:27, 13 April 2020

Agenda and Minutes

GOC Meeting - May 2020

  • Paris consortium meeting scheduled for May 12th - 14th will be held virtually
  • Please keep the dates open
  • More details will be forthcoming, but note we are planning for shorter days to accommodate all time zones as best as possible
  • Agenda
  • Users meeting - May 11th

Review of annotations to GO terms that refer to ChEBI roles (Karen)

  • Referring to ChEBI roles in logical definitions for terms in the GO leads to incorrect inferences, so we will remove references to roles
  • Terms that refer to specific chemicals will be classified in GO according to ChEBI's chemical classification, just not according to a, perhaps context-specific, role
  • If needed, curators can request new chemical-specific terms, but should consider if these chemicals are endogenous to the organism or synthetic, and how many annotations would actually be made to that term
  • Please review annotations that refer to these types of terms; annotation review tickets links to ontology tickets with more explanation of the issue
  • cofactor & coenzyme roles
  • drug & xenobiotic roles
    • drug terms will be obsoleted in favor of xenobiotic terms, as at least a xenobiotic can be defined wrt the organism whose gene product(s) are being annotated
    • long discussion of drug roles in GO here

File Formats

Proposed GAF Update, GAF 2.2

  • GAF 2.1 will continue to be produced and supported by the GOC into the foreseeable future
  • We are proposing an incremental update to the GAF, though, to allow for use of the full set of gp2term relations
  • Proposed set of allowed relations is in this github ticket: https://github.com/geneontology/go-annotation/issues/2917
    • This would be the same set of gp2term relations used in the GPAD, the main difference is that they are in the same column as negation and when both apply they are pipe-separted
    • Default gp2term relations:
      • Molecular Function: enables
      • Cellular Component: located_in
      • Biological Process: individual groups decide based on annotation practice
  • Why do this?
    • The expanded set of gp2term relations is available in annotation tools, e.g. Noctua and Protein2GO, but by not including them in the GAF, we don't give the GOC, or users, any mechanism to filter specific sets of annotations for GAFs and subsequent analyses. Making this change will allow GO and groups to do this.

GPAD/GPI 2.0 Specifications

  • Questions/Comments
    • We need guidelines for how to represent proteins ids or accessions shared by multiple gene products, e.g. histones. [Stacia]
  • Response:
    • Groups should work with UniProt or internally to disambiguate protein records that refer to multiple genes, i.e. generate gene-centric protein entries in UniProt
    • In the interim, we are proposing that protein entries that correspond to the product of multiple genes be included as a single entry in the gpi, with one name chosen as a symbol, and other names listed as synonyms. All parent gene ids will be listed in the 'Encoded_By' field.

Deadlines

  • Groups have asked by what date they need to produce these new files.
    • Alliance members that will be migrating to Noctua, will need to use the gpad/gpi 2.0 file format for the data ingest.
    • Otherwise, GO will continue to consume and produce GAF for the foreseeable future, but note that gpad/gpi is a richer and more robust file format and is preferred for data exchange within the GOC.
    • For GAF 2.2, the software group will need some time to prepare the pipeline and other software (e.g. AmiGO) and we will also want to announce this change to the community, so there will be some lead time here (exact TBD).

Use of Relations in Annotation

  • Consistency in use of gene-product-to-term relations is an ongoing concern
  • Previous conference call we discussed soliciting ideas from groups about how to leverage expertise in specific areas of biology or curation to help achieve consistent representation in our annotations
  • Follow-up
    • In what areas of the BP branch do curators have expertise?
      • Also look at Annual Reviews? Organism-specific 'book's, e.g. WormBook?
    • Discuss in more detail at May meeting to begin formulating a plan and more concrete deliverables, e.g. GO-CAM templates, curation guidelines?

Attendance

  • On call: Chris, Colin, David, Dustin, Giulia, Harold, Helen, Karen, Kimberly, Laurent-Philippe, Pascale, Petra, Rob, Sabrina, Seth, Stacia, Suzi