LEGO September 26, 2016

From GO Wiki
Jump to: navigation, search



Training and Documentation

USC Training Session

  • It will be held 2016-11-07, the Monday after the consortium meeting
    • We will probably need to do another session as well. Zfin will probably want a session on site and the SGD curators may be able to piggyback on that session. Need to think about dates for a Zfin session?
  • Textpresso-Noctua summit at Caltech - Thursday, November 3rd or Tuesday, November 8th?


  • Some comments and edits made to the Quick Start guide
  • Any more comments on Quick Start Guide?
    • The new quickstart page is linked from the web site now.
    • We should change the learn more button to something else. It is not clear exactly where it will lead. It could simply be an 'about the project' page. We need to make it clear that it is instructions.
  • We would like to make an FAQ.

Software Updates

Updates to UI

NEO Overview and GPI Files

  • Questions, issues still to be sorted out?
        • Chris will present his GPI talk at the GOC meeting.
    • We have entries for:
      • Genes
      • Proteins
      • Transcripts
      • ncRNAs
      • Protein Complexes
    • How should protein complexes be represented?
  • David will modify the wiki that describes the GPI specs.

GAF/GPAD Outputs

  • Set up time for meetings on this.
  • The GPAD output is still creating files with RO identifiers in them. They need to be translated to the relation names.

See model-57c82fad00000358

MGI Meeting Follow Up

  • Review the list of software and annotation issues that were discussed at the MGI training session, June 15th-16th.
  • See the Google doc
  • Some specific follow-up:
    • GAF/GPAD output is probably highest priority
      • Remaining issues:
        • How to handle causal chains
        • Multiple evidence = multiple lines in the GAF
    • We still need to figure out attribution. Right now everything coming in from Noctua has the group attribute GO_Noctua. That means that different groups making annotations to mouse genes cannot be distinguished.

LEGO Relations

  • New relations needed:
    • removes_input
      • directionality for causally_upstream_of_or_within and causally_upstream_of
        • GH tickets have been created to request these relations.


  • On call: Barbara, Chris, David H, David OS, Guilia, Helen, Jim, Kimberly, Midori, Paul, Ruth, Sabrina, Seth, Stacia, Suzi
  • Meetings:
    • USC - Monday, 11/7, training session
    • Noctua/Textpresso - developers meeting at Caltech, Thursday, 11/3


  • Quick Start Guide
  • FAQs
    • Send questions to include

Software Updates


  • Noctua will be down a bit later today; Seth will send an email
  • Working on some clean up

gpi file

  • Chris will give talk at USC meeting about creating NEO and gpi specs
  • Not getting gpi files from many MODs yet
    • There is a parallel project for this in AGR (the AGR portal)
    • Would the various MINE (e.g. FlyMine, WormMine) work for getting the gpi files?
    • Some efficiences could be gained here.

AI:Curators should learn what the development state of their respective Mines is and whether additional work needs to be done to get all of the relevant information for a gpi from the Mine instance.

  • MGI's gpi: genes, proteins (PRO ids), transcripts, ncRNAs
    • RNAcentral is only curating ncRNAs, not transcripts
    • MODs could use transcript ids if they have them, otherwise we should standardize on say, ENSEMBL IDs
    • No protein complexes yet in MGI file
    • Will have an in-depth discussion of protein complexes at USC meeting
    • You can make complexes on-the-fly in Noctua (see Noctua VIMEO channel)
    • What relation to use between complex and protein members?
      • has_part or part_of
      • Has implications for folding
      • Want to use has_part - need to include this in the documentation


  • Will have a separate call for this - David H, David OS, Jim, Kimberly - 10am Wednesdays, starting this week; can use the same Bluejeans channel
  • What is the consensus for what legacy annotations should be created?
  • If the annotation is in the model, it will be included in the GAF/GPAD
  • Any objections to that approach?
  • On the next call, we could specifically address this approach with examples.
  • We need documentation and examples for this.
    • For example, see Alan Bridge's cholesterol/dauer model or other models that may include what might be thought of as downstream processes.
  • Need to discuss this at the larger GO consortium meeting - this is a BIG issue!
  • We also need to think about how we can migrate legacy annotations, e.g. IDA annotations may be amenable to part_of while IMP annotations may be much more of a mix of causally upstream of, regulates, part of.
  • Any plans for a tabulated display that would see exactly what annotations would be exported in a GAF/GPAD file?
    • Yes, there are plans for this, but exactly how this is going to look is not quite clear yet.
  • At some point, we need to start thinking about the exit plan from GAF/GPAD.
    • Users may still want/need to use GAF/GPAD.
  • We need a MOD-wide plan for consuming the OWL models - this will be a collaborative exercise between the MODs and GO.
  • There are a number of things mentioned in the grant that are good starting points for this discussion.
  • What about GAF files for enrichment analysis?
    • These tools will still need a GAF for now, but the future is in the OWL representation.
  • Are there existing tools that can use OWL?
    • There are tools that use the topology of the graph, and these types of tools could be reused. Note that these tools are only using a subset of the information.
    • LEGO is new, so no one is using the information yet.
    • Enrichment analysis can be done using Reactome and KEGG; display of enrichment results on pathways is informative.
    • We'd want to display results on LEGO models.
    • Should Peter demo this at the USC meeting? Ruth?
    • Panther has been working on integrating Reactome data into enrichment analysis.
    • GAF/GPAD will not be completely taken away, though; that's why we're having these discussions about creating GAF/GPAD.

LEGO Relations

  • David OS has asked for the necessary relations, i.e., the directionality for causally upstream of relations
  • New, improved interface will make it easier to choose the correct relation, as there will be context-sensitive menus for this

Labeling and Editing Models

  • Can we have more or different labels for models?
  • In particular, is development not suitable?
  • Review is the middle state between development and production
  • Production is ready-to-go
  • We talked about the idea of locking a model, as curators may want to be able to keep a particular model unchanged
  • This also speaks to the annotation dispute idea/functionality
  • Is there history of models? Yes.
  • This is an important issue and we need to discuss this further.
  • Do we want moderators for models? Moderators would be the curators who specify when models would go into production.
  • We want to have at least two sets of eyes looking at and reviewing a model.