GO-CAM Noctua Call 2018-06-13

Meeting URL

https://stanford.zoom.us/j/679970729

Agenda

Internals

• From 5/23 meeting:
• refactoring
• stack alignment
  • common widgets/UI
• paying down debt
• functional parity
  • live updates
  • login and groups across clients and workbenches

Annotation

• If there is plan to just have an update minerva API, preferably batch update just like currently adding multiple entities at the same time. Altering the individual class expression on that low level is very error-prone. For example I just found out, changing the term and evidence is 2 separate processes. (Tremayne)
• Editing ('conventional' and GO-CAM)
  • Tremayne working on an editing interface
  • Curators can search for gene/gene product ID, GO term, Reference
  • Matches will then be presented in a form view for editing with links to a model, if applicable
• Model cloning

Error checks

• The Form to use Reasoner or centralised checker through API to check for errors. I think Noctua Form error checking was for 1.0 release. It depends heavily on golr and different way than Reasoner. I think we should have one definition of an model activity (algorithm wise). Bcz there will be cases where the graph says there is an error and Form doesn't catch it then. (Tremayne)
• Incorporating annotation rules into Noctua
  • Some annotation errors are flagged in the form editor
  • Others may need to get checked during the release pipeline
    • Curators may wish use lines from GPAD file in order to retrieve the correct model id if they want to edit starting from a model URL

Evidence codes

• Working with ECO to create manual/automatic assertion children

GPAD Outputs

• Protein complexes
  • This will need to be one of the high priority tasks for the working group (see below)
  • Emergent functions will need to be handled differently from functions where one of the subunits actually enables the activity of the complex
• Annotation extension relations white list
  • How do we want to handle this?
  • Do groups want to see all possible extensions and then filter at their respective MODs?
  • If we do that, are there any possibilities for inconsistent display between MODs and AmiGO?

Landing Page

• I have a particularly blocking issue with NEO and URIs that prevents me to correctly display the name of genes when browsing the models. (Laurent-Philippe)
• Model searching (production)
• Model titles - need guidelines
  • This issue is still outstanding
    • We may need different guidelines for different types of models

Model Seeding

• From GAFs
• From Reactome

Relations

• Reconcile CC annotation relations with CC ontology relations
• Use of has_input vs has_participant for binding annotations
• David, Pascale, and Kimberly reviewing all relations used in GO and their documentation

Templates

• Create and use templates for curation

Text Mining

• TPC integration into form
  • Will draw up requirements for this and hope to make progress before and/or during the late August hackathon at USC

GO-CAM Modeling Working Group

• Tuesdays at 8am PDT
• Proposed GO annotation meeting schedule:
  • 1st Tuesday: Alliance Gene Function
  • 2nd Tuesday: GO Consortium
  • 3rd Tuesday: Alliance Gene Function/GO-CAM Working Group
  • 4th Tuesday: GO-CAM Working Group
  • 5th Tuesday: ad hoc, as needed
• One Zoom URL for all: https://stanford.zoom.us/j/976175422

Minutes

• On call: Ben, David, Dmitry. Dustin, Harold, Jim, Karen, Kimberly, Laurent-Philippe, Liz, Paul T., Peter, Rob, Sabrina, Sandy, Suzi A, Suzi L, Tremayne

Annotation

• If there is plan to just have an update minerva API, preferably batch update just like currently adding multiple entities at the same time. Altering the individual class expression on that low level is very error-prone. For example I just found out, changing the term and evidence is 2 separate processes. (Tremayne)
  • Tremayne and Jim will discuss further to possibly modify the Minerva API to facilitate the operations that need to be performed to update annotations.

Error checks

• The Form to use Reasoner or centralised checker through API to check for errors. I think Noctua Form error checking was for 1.0 release. It depends heavily on golr and different way than Reasoner. I think we should have one definition of an model activity (algorithm wise). Bcz there will be cases where the graph says there is an error and Form doesn't catch it then. (Tremayne)
  • Tremayne and Jim will also consult on this to try to develop a consistent way to check for errors. This is particularly relevant for traversing from the graph editor to the annotation summary table as well as for copying from the summary table back to the form for quick editing.

GPAD Outputs

• Protein complexes
  • High priority for GO-CAM working group to sort out exactly how the different flavors of protein complexes and the activities of the complexes and respective members should be annotated in Noctua/GO-CAMs and output in GPAD
  • This is an area where coordination with PRO would be good. For example:
    • When a curator creates a new complex (e.g. protein complex has_part A, has_part B..) that could initiate a term-genie like request to PRO complex portal for a new identifier. When a complex already exists, it returns the id.
• Annotation extensions
  • Another mandate for the GO-CAM working group
  • How many and what kinds of annotation extensions should be output in GPAD?
  • On a related note, and also relevant to error checks, there are still some combinations of annotation extension relations and values that violate Domain/Range constraints
  • We need to integrate these error checks at some point in the Noctua curation pipeline, preferably as early in the curation process as possible

Landing Page

• I have a particularly blocking issue with NEO and URIs that prevents me to correctly display the name of genes when browsing the models. (Laurent-Philippe)
• Model searching
  • Would be good to search on development as well as production models
  • A read-only SPARQL service might be helpful for implementing this