Difference between revisions of "2017 Cambridge GOC Meeting Agenda"

From GO Wiki
Jump to: navigation, search
(Tuesday 3rd October)
Line 106: Line 106:
===Tuesday 3rd October===
===Tuesday 3rd October===
====Centralization of InterPro2GO annotations====
Proposal (follow-up from Geneva 2016): (Paul T)
*GO database pulls directly from InterPro2GO for UniProt Reference Proteomes
**MOD identifier is used as primary gene identifier
*Annotations are given "contributed by" InterPro
*MODs pull from GO database, no need to maintain separate InterPro pipelines
Line 112: Line 119:
* GO ribbon  
* GO ribbon  
Presentation by Mary + 30 minutes discussion
Presentation by Mary + 30 minutes discussion
===Wednesday 4th October===
===Wednesday 4th October===

Revision as of 10:33, 25 September 2017

GOC Meeting, Cambridge , October 2-4, 2017


Monday 2nd October

Working group meetings

  • QC team brainstorming (Kimberly, Sylvain, Val, Pascale)
  • Noctua (Kimberly, Seth, Chris, David H, Paul, Karen, ...)

Annotation issues

  • Report from signaling workshop: David /Kimberly
  • Signaling: First attempt at Annotation consistency 2.0 - Kimberly to report on the approach and the outcome.

Discussion points: Limited participation (self-selected to participate). One recommendation may be to ask all active curators to participate, even at some low level https://github.com/orgs/geneontology/projects/8

Annotation guidance presentation

Pascale/Sylvain + move slides to Google drive

Representing biologists’ view of biology

Pascale /Ruth author intent & protein domains; IDA v IC v New evidence code https://github.com/geneontology/go-annotation/issues/1621 30 minutes discussion about issues; hopefully action items can come out of the discussion

CC component annotation guidelines


    • CC component annotation: what does it mean ? Kimberly to do 1 proposal (out of 3 alternatives)

1. where the protein is active 2. two different meanings: enables or the right RO (part:of, ie just found there) 3. part_of (low information value!)

Contributes_to guidelines


HTP guidelines

Helen - 15 minutes Guidelines draft Action Items Corvallis 2017/0

  • add to guidance: contact authors if going to add - I think this should be done for all papers (Ruth)
  • add to guidance: mark paper that is not annotable in Protein2GO (or your curation tool). Also consider contacting authors/journal to inform them that not going to be annotated. But if the authors have fully explained the problems then it is not appropriate to contact the authors/journals as they know the limits of the approach (Ruth).
  • add to guidance: HTP IPI data should not be submitted to GO, instead send this data to IntAct. (Note:Done)
  • change evidence codes for legacy HTP annotations to HDA, HMP etc.
Transcription annotations decision tree

Ruth Action Items Corvallis 2017/06

  • David OS to create transcription regulator activity (proposed by Paul T)
  • Proposed changes to decision tree in Corvallis:
    • Simplified from previous version. Essentially a choice between ‘regulating transcription by RNA polymerase II’ or ‘regulating gene expression’
    • Annotation 5 = contributes_to sequence-specific DNA binding

David: when people do enrichment, they don’t drop contributes_to (ie., pay attention to qualifiers), so all those proteins will come down as ‘DNA binding’ Action item: replace ‘annotation 5’ with ISS annotation (if DNA binding domain) or contributes to annotation 5 with ISS annotation if no DNA binding domain and domains to suggest coactivator

    • Annotation 3 = nuclear chromatin

Not everyone comfortable with this (ex., Stacia, David) Shouldn’t it be ‘colocalizes_with’? It was previously agreed that the definition for nuclear chromatin: The ordered and organized complex of DNA, protein, and sometimes RNA, that forms the chromosome in the nucleus. Source:PMID:20404130 was to be applied. The proteins here include all associated proteins, not limited to just histones. With this statement the TFs are then contributing to chromatin, rather than binding to chromatin or colocalizing with it.

    • Decision tree to be put up on GOC website.
    • Should annotation 4 exist? (Ruth)

Use of Qualifiers in Legacy Annotations

Pascale Proposal: We will apply the general qualifier to all legacy annotations. Each group can provide more specific qualifiers if they have a mechanism to distinguish. Action Items Corvallis 2017/06

  • Working group to decide what relations should be available in Protein2GO or other tools for manual annotations. Also decide when there are different ways of expressing the same thing, what way we will choose. What should the default gene/gene product relation be for legacy annotations?
  • Chris to work on reports that may help curators make decisions about what annotations can get more expressive qualifiers.

Multiple qualifiers for an annotation


Expanded list of relations


Regulates relations

Kimberly Adding new qualifiers for the relation between a gene/gene product and a GO term What should the default relation be? How will we handle regulation? Use a relation, involved in regulation of, or use the precomposed regulation term?

Ruth: There are now 3 ways to say the same thing: - involved_in_regulation_of X - involved_in X regulation - involved_in BP regulates(X)

For annotation purposes and for our users we want one.

DOS: Good point. These are semantically identical, but I agree we need to find a way to only have one: by convention for classic GO annotation and by filtering the output of inference for noctua output. Proposal: If a named regulation class exists: involved_in X regulation ...if not: involved_in {some BP} regulates(X) NOTE: If there was an annotation in Noctua such as ‘regulation of’ ‘very specific term’ and there was no term as ‘regulation of very specific term’ the GOC pipeline would create the annotation to the parent term: ‘regulation of less specific term’. Q: Is this implemented ?


Establish a working group for when and how to use proteoforms in annotation. (Kimberly with Harold and Karen)

  • Evening: Poster session

Tuesday 3rd October

Centralization of InterPro2GO annotations

Proposal (follow-up from Geneva 2016): (Paul T)

  • GO database pulls directly from InterPro2GO for UniProt Reference Proteomes
    • MOD identifier is used as primary gene identifier
  • Annotations are given "contributed by" InterPro
  • MODs pull from GO database, no need to maintain separate InterPro pipelines


Philosophies and methodologies (Mary, Suzi)

  • Should GO support one or many? Creating a biologically useful slim (complete coverage by aspect, biologically useful terms i.e sufficient granularity, avoiding single step process terms (i.e functions), different slims for different purposes: Judy, Mary D (+ Suzi, Val, etc)
  • GO ribbon

Presentation by Mary + 30 minutes discussion

Wednesday 4th October

AM Workshops

  • Priorities:*


Hands-on/Workshop Suggestions

  • Displaying GO annotations (non redundantly, in MODs, in AmiGO), general data presentation.
  • Good annotation practice: phenotype vs process, for eg.
  • Feedback for QuickGO beta (small group session with Sangya, 1h)
  • Making tickets/revising annotations/overflow discussions from the signalling workshop session

GOC meeting Discussion topics

  • Sets of relevant genes for human diseases: (Judy - analysis report)
  • Funding opportunities for targeted annotation...
    • human disease specific
    • via DoE or NSF for microbes?
    • metagenomics?
    • getting B. subtilis from Peter Karp?

Project updates

  • AGR - report to GOC: PIs
  • SynGO meeting report: Paul T
  • Report of Reactome-GO connection: David H/Peter D
  • Report on transcription work/Noctua templates: Astrid GREEKC consortium


Follow up with Seth

(Pascale Q: My understanding is that there are essentially no resources for AmiGO right now so any AmiGO issue is low priority. Is this correct?) (Seth: Pretty much--all current work at this point is basically "hobby time")


Huaiyu Action Items Corvallis 2017/06:

  • Encourage discussion between PAINT curators and other annotators about terms not used for propagation
  • Report how many annotations per species are used for annotation propagation; could even supply this number for propagation specifically to human genes
  • Ruth and Huaiyu (others?) will discuss making use of groups that have already annotated specific gene lists to annotate the corresponding PAINT families.
  • Smooth out the challenge mechanism to make it easier to do make and resolve

the challenges, identify terms that may be problematic and would benefit from consistency exercises and discussion.

  • Get a list of families where terms have not been propagated (?) - Please check this one for clarity.
  • (added to github project board) Develop mechanism to trigger review of annotated PAINT families.


Kimberly Action Items Corvallis 2017/06

  • Continue with the consolidation of all documentation for ontology editing, and remove all old documents.
  • Review annotation documentation and add to github and readthedocs.
  • Make sure to mark obsolete pages/doc as such and add a link to the new relevant doc
  • Solicit annotation documentation from participating groups for consolidation.
  • Make sure to mark docs with ‘Date last reviewed’ so it’s easy for users to know when the documentation was last touched.
  • Pascale: additional information associated with GO terms, see GitHub ticket?
  • Add and follow action items at https://github.com/orgs/geneontology/projects/3

Action items & tickets from previous period


Getting Noctua ready for production

Kimberly & Seth Blocking issues list: TO BE COMPLETED

Action Item Corvallis 2017/06:

  • Provide ways for users to recover and digest GO-CAM units (Gene Ontology-based Causal Activity Model). Ideas include rule-based generations of text statements from model, cytoscape view of network described, etc.
  • ECO codes available for use in Noctua should show how they map up to a classic GO code, and there should be an alert for curators when they are using a code that does NOT map up to a classic code
  • PRO IDs for use in Noctua
  • Fix GPAD export from Noctua https://github.com/geneontology/noctua/issues/418
  • Add a SPARTA workbench https://github.com/geneontology/noctua/issues/465

Action Item Corvallis 2017/06:

  • Working group discussion of evidence on complex Noctua models (Kimberly?)

Noctua table view demo


GitHub tutorial

(Seth & Pascale): how GO now uses GitHub for project management and guidelines for contributors

Update on MF refactoring (Pascale or PaulT)

Action Items Corvallis 2017/06

BP refactoring
  • Defining “Cellular Process” and “Multi-Organism Process” terms
  • Action item: the comments/examples/notes should be captured, working group to discuss this.

The Fate of Simple Processes
  • analysis of gene products annotated to phosphorylation but not annotated to a kinase MF term. What did these annotations actually mean? Curators would need to review. Timeline? Deadline? Working group? One possibility would be to use part_of/involved_in relation to phosphorylation for genes also annotated to kinase MF, and causally_upstream_of_or_within for others until curators can re-evaluate.
  • Other considerations:
    • Helen: are there many the single-step processes?
    • Paul: user-oriented approach - is it a useful grouping for our users?
    • Ruth: we need to think about the meaning beyond just “phosphorylation”
    • Ruth: what are the consequences of removing “phosphorylation” and what happens to all its children?
    • Pascale Q: providing that we remove the processes, would it affect the term enrichment analysis?

Viral processes

Pascale Action Items Corvallis 2017/06 Check whether the incorporated changes could affect the host proteins. Document the use cases.

Usability issues

  • What do users want?
    • truth (summary/story/model)?
    • fishing expeditions?
    • api access?
  • Who are our users?
    • geneticists?
    • clinicians?
    • computational biologists?
  • Use cases: https://github.com/geneontology/go-ontology/issues/13606
  • Perhaps discussion of examples of how the GOC members are engaging with the community and how we can do better in the future (suggested by Helen).

Lower priority

Citing GO

Action Item Corvallis 2017/06:

Postponed to next GOC meeting

Quality control

Action Item Corvallis 2017/06:

  • Explore implementing periodic reanalysis of published gene sets and making this available to users to illustrate how the dynamic nature of GO (should) improve analysis. P. Pavlidis’ http://gotrack.msl.ubc.ca/ may be particularly useful here.