GO-CAM Conference Call - 2019-04-23

From GO Wiki
Jump to: navigation, search

Call Information

  • See GO's Google calendar for Zoom URL
  • email Kimberly if you need access to the GO's Google calendar

Agenda

Follow-up from Cambridge GOC Meeting

Gitter for go-annotation

  • Reminder that we will be trying to use gitter as an informal chat room for GO annotation questions
  • https://gitter.im/geneontology/go-annotation
    • login with your github account
  • This doesn't replace github/go-annotation or formal documentation on the wiki, but is meant to be a place to ask questions or share ideas
  • We will monitor the conversations to make sure that key issues are elevated appropriately

Annotation Imports

Naming Models

  • Follow-up from breakout discussion at the Cambridge meeting
  • Model titles are currently visible on the Noctua landing page as well as on the public GO-CAM site for production models only
  • Current naming practice:
    • MOD imports - by MOD gene ID, e.g. WB:WBGene00004732 or MGI:MGI:101757
    • SynGO imports - gene/gene product name, GO aspect, number (?)
    • Manual models:
      • Production, Development (in progress, test)
      • No convention - gene names, species, pmids, mod paper ids, curator initials, paper titles, biological processes, model ids, etc., or combinations of these
        • The diverse titles arose partly out of a desire for curators to easily find production and development models they'd made
        • Do we all agree that this is not optimal and standardized model titles is desirable?
  • Do models need informative, human readable titles?
    • If yes, how should they be generated?
      • Proposal from Cambridge meeting:
        • Use controlled vocabularies and lists to generate consistent model titles
          • Manual or auto-generated?
          • Current tool behavior - a model can't be saved unless it has a title
          • Could a temporary title (e.g. model id) be generated for saving and then curators could construct a model title from a drop-down list of terms and entities present in the completed model?
            • Key points:
              • Aiming for organism-specific models of biological processes - BP terms would thus get preference for naming?
                • Other key information: anatomy and/or cell type, life stage, species
              • Annotations not yet part of a larger process would be added to the gene-specific model?
  • What should curator workflow be when starting to annotate?
    • Search existing models for possible matches?
      • This requires searches on development, as well as production, models

New Noctua "Workbenches"

Noctua Form

Relations Editor

Annotation Review Tool

Minutes

  • On call: David, Dustin, Edith, Harold, Karen, Kevin, Kimberly, Marie-Claire, Monika, Paul T., Petra, Rob, Sabrina, Seth, Stacia, Tanya, Tremayne

Gitter for go-annotation

  • Chat room available for GO curators to ask questions and share ideas
  • Curators are encouraged to login and participate

Annotation Imports

  • Dustin has begun to implement rules for importing 'has regulation target' annotation extensions
  • David and Kimberly are QCing the resulting models for accuracy
  • 'has regulation target' annotation extensions were not always clear, so if we can establish better rules for capturing targets of regulatory processes, then we can implement those into conventional annotations as well
  • Use of the chained relations, e.g. regulates_o_enabled_by, may ultimately be better for clearly capturing the object of regulation

Naming Models

  • We discussed considerations for naming models
    • There are currently three different types of model names: MOD imports (gene ids), SynGO imports (gene, aspect, internal id), and manually assigned titles
  • Does it matter what we name them?
    • Right now, models names are used, in part, to search for models that curators have made
    • Implementation of better search interfaces for curators should alleviate the need to put model metadata in the model title
    • Users and curators may have different needs, so perhaps the name of a model may ultimately matter more to GO users who are browsing models than to curators?
    • The Annotation Review Tool provides much more search functionality for models than is currently available and also allows editing within that interface. We will demo this on a future GO-CAM call.
    • One good suggestion from the Cambridge meeting was to name models as if you were composing a title for a review article.
  • Can we just use machine-generated titles?
    • Maybe, but we'd need to explore how these would be created.
    • Machine-generated titles would remove the naming burden from curators, but we'd still like to have informative model titles for browsing.
  • We will follow-up with individual groups to determine what are some of the barriers to using Noctua
    • Incorporation of real-time annotation rules and error reports are two things curators really like to have
    • We need to make sure that the current work-flow at each of the MODs can still be achieved using Noctua, in addition to having the capabilities of creating more expressive GO annotations