Manager Call 2017-02-15: Difference between revisions

From GO Wiki
Jump to navigation Jump to search
No edit summary
No edit summary
Line 35: Line 35:
GOC Meeting:
GOC Meeting:
* Reactome workshop details are necessary.  
* Reactome workshop details are necessary.  
** ToDo: Suzi to contact Pankaj to get more details about it and distribute them.  
** '''ACTION''': Suzi will contact Pankaj to get more details about the workshop, and will distribute these details to consortium.  





Revision as of 13:13, 15 February 2017

Bluejeans URL

Agenda

GOC Meeting in June

  • Curators have asked for more information about Reactome workshop in order to make their travel plans
  • Who will contact Pankaj and Peter for that and report back to the group?

Protege Workshop in Berkeley Next Week

  • What else needs to be done for prep/planning?

Annotation Conference Call

  • Some big issues facing the annotation group
  • Addition of new qualifiers/relations was thought to be desired and important for several annotation streams - conventional, PAINT, LEGO
  • However, the questions of which ones to use and how to apply the ones we do use consistently are still open
  • We need a good way to coordinate annotation work to make sure we actually get some useful guidelines
  • One suggestion - divide up curators to look at specific areas of biology to generate guidelines for that area
    • There's a lot of value in what Val has done for the cell cycle, but would guidelines for cell cycle apply to development, behavior, metabolic processes?
      • For example, what is the relation between a transcription factor required for specification of a particular neuron, and the behaviors mediated by that neuron?
    • PAINT view of annotations can be very informative here
    • LEGO models will help here as well
  • Other possibilities - examine genes that are annotated to both a process and regulation of that process
  • Are the autophagy and apoptosis projects models for this work?
  • Should part of this work also include a systematic analysis of what changes result from using different annotation sets in enrichment analysis?
  • How much do we need to coordinate this work with ontology development, specifically the molecular function refactoring and design patterns?
  • Can we come up with a plan and propose it to the consortium?

Minutes

Present on the call: Kimberly, Moni, Melanie, Paola, Suzi, DavidH, Pascale, Huaiyu, and Chris.


GOC Meeting:

  • Reactome workshop details are necessary.
    • ACTION: Suzi will contact Pankaj to get more details about the workshop, and will distribute these details to consortium.


Planning for the Protege workshop at Berkeley:

  • Chris sent around an email with requirements and installs necessary for the workshop.
  • GitHub client to use: the one on the command line - everyone must use the same client, discourage the use of UI Git clients.
  • Protege
  • Documentation is scattered but please go through it to prepare for the workshop
  • David intends to catenate and compress the documentation.
  • The idea for the workshop is to teach the basic actions to take in order to edit the ontology. Also to collect knowledge on ‘frequently used’ actions.


Annotation Conference Call:

  • Challenge: Speaking broadly, the efforts of the annotation working group could be a little more focused. For example, while discussing the proposal to add qualifiers to the GPAD and GAF, it was evident that there seems to be confusion regarding what these relations mean biologically. This is causing a delay in adding the qualifiers until curators are more clear about whether or not to use them and when.
  • Suggested solution: Let’s put “proto” documentation together about the reasoning the GOC (us) is proposing to use, and have the community comment on that. Pascale will begin putting together guidelines on how to use the proposed qualifiers. Estimated timeline: it will take her about 2 days to do this. She will work on it the week after the Protege meeting. Will loop in David and Kimberly for added commentary.


ACTIONS:

  1. Pascale will look at IMP rules and Val’s matrices, paying close attention at ambiguities.
  2. Chris and David will work on enrichment analysis with and without IMPs included.
  3. Kimberly will look at GO and phenotype enrichment.


WHO is responsible and HOW they will go about these ACTION items:

  • Start with a review of IMP + Regulation + Process; then look at the evidence to establish whether or not this is an overstatement. (Pascale)
  • Take a look at Val’s matrix analyses, which identify ‘causally upstream of’ relations. (Pascale)
  • Review also issues discussed at https://github.com/geneontology/go-annotation/issues/1532 (Everyone)
  • Modeling what would happen: Run an example enrichment analyses where all the IMPs are removed? It can be done using Vlad. See the tool at http://proto.informatics.jax.org/prototypes/vlad/ It could also be easily done via AmiGO, using a script (connecting to PANTHER). This test should be done on a published set. (Chris and David).
  • We should also take a side-by-side look at GO and phenotype enrichment, allowing us to inspect IMPs. Using phenotype enrichment tool. (Kimberly).


Additional considerations:

    • We must keep in mind: what did the original annotation mean?
    • We will take a few examples from every group, and explore the idea of having an initial default qualifier, e.g. causally_upstream_of. Can we salvage some of the original annotations to show that they are indeed expressing the appropriate statement? Transcription factors are a good place to start.
    • Conventional annotations some time just don’t express all we want them to. Working on these qualifiers will help us increase that expressivity.
    • Important to study how to retrofit the current corpus of annotations.
    • Important to consider whether we want to ‘trash’ annotations that are causally_upstream_of. LEGO gives you the causal chains. But to make a single annotation to causally_upstream_of when conducting gene enrichment analyses, a lot of noise will be created.
    • Huaiyu: these annotations will be useful. With a qualifier, they may restrict the noise when conducting enrichment analyses.
    • The IMP evidence does not prove that the annotation is true, but rather supports that the annotation is true, and curators use their judgement to assess the validity of the relationship.


How much do we need to coordinate this work with ontology development, specifically the molecular function refactoring and design patterns?

  • Need to come up with a plan and propose it to the consortium. Pascale enthusiastically suggested we can do this.