Manager Call 2018-11-21: Difference between revisions

From GO Wiki
Jump to navigation Jump to search
mNo edit summary
 
(17 intermediate revisions by 2 users not shown)
Line 3: Line 3:
= Agenda =
= Agenda =


== Montreal GOC Meeting Action Items and priorities==


===Big picture/projects===
==Annual report==
What are our main priorities? What is the strategy to achieve them?
What do we need from MODs ? We have been asking staff, personnel changes, publications, curation statistics from the current year. Looks like we have only included publications and overall number of publication curated in the last report. 
How about asking:
* Number of FTEs on the project ?
* Publications and presentations
* Any highlights they would like to share in the annual report


Kimberly, David, Pascale: Proposal for 2018/early 2019 priorities:
** Do we need any additional information ?


* '''GO website migration''': Suzi L/Laurent-Philippe: Aim: Dec 1
==Follow up on priorities==
** Requires many technical tasks that mostly require Seth
* '''Release''': came out on Nov 16:  
* '''Data release pipeline''': Pascale/Seth: Are there any blocking issues remaining? If not, we can de-prioritize this for now
** Suzi A, will you announce ? Do we have a process?  
** Many steps are inefficient and slow
** Why was it delayed? Do we need more resources on releases ?  
** More work needs to be done: Most work is done by Eric right now - can anyone else help on that ? Paul can find more resources - what do we need ?
** Release tasks update (Seth)  
** In the longer term: We need a person who will be responsible for the pipeline - possibly 25% (long-term estimate)
* '''GO website migration''': Suzi L/Laurent-Philippe: Aim: Dec 1 -> Is this still the plan ?  
* '''Noctua+GO-CAM repository''': Kimberly/Seth/David:  
* '''Noctua+GO-CAM repository''': Moving legacy annotations to GO central: Jim, Ben, Chris, Seth: strategy
** A developer is needed full time to develop until the Cambridge meeting:
** Tasks: (just a teaser): 
*** getting in all existing annotations in - including implementing gp2term relations implemented
**** Chris: We will not be able to bring in everyone's annotations in by Cambridge
**** Back end: how much work will that be? Issues, we need to figure out how to partition annotations. We have some research work about this, there are many possible ways. We need to decide on the strategy.
**** Paul: is there a solution where nothing is connected at first (ie, 1 model for every annotation)? This would avoid create dependencies on connections AT FIRST. Advantage is that every group could move to Noctua as soon as that's in place. Otherwise we will likely be delayed.
**** David: we need to have a plan for how models/individual annotations will be handled/re-merged
**** Kimberly: AI we should make a Google doc - TO DISCUSS at the next managers call (?)
**** Pascale: we also need to think about doing QC on the automatic tools that would predict connectivity - that will take quite some time too, so prefers a more simple solution
**** Chris suggests doing an experiment in which we test integrating unconnected annotations, for example starting with dictyBase
**** Proposed import process: take triples out of each model
*** improve RO – who should do that? Including testing property chains:
**** Chris: We can have a bit of Jim's time on this.
**** David: What will we do about the different 'participants' relations - are we just moving forward ? Chris: yes
*** work on reasoning
*** fixing issues with pipeline (including GPAD output)
**** required GPAD2, which will require some plumbing and will break for a lot of people: people importing GPAD 
**[https://docs.google.com/document/d/1FSWy7mvsSN7yZ10TL5CzSeJrzverub_oYwHY25bKD9o/edit Noctua Roadmap]
* '''GO rules (includes QC)''': Pascale/Eric: Improving reports: end of 2018, implement/fix some rules 4 rules /month?
* '''Tools''':  
** GO API implementation (Laurent-Philippe / Deepak)
** Bioinformatics Guide (how to):
*** Ontology manipulation / traversal (parents, children, slim, related terms, etc)
*** Search / Browse / Aggregate annotations (e.g. per gene, taxon, pathway, aspect, etc)
*** Interactive Online Notebooks:
**** Jupyter (Python)
**** R Markdown (R)


Projects:
* '''Data release pipeline''': Pascale/Seth
* '''GO rules (includes QC)''': Pascale/Eric
* '''GO website migration''': Suzi L/Laurent-Philippe
* '''Noctua''': Kimberly/Seth
* '''GO-CAM integration with other pathway resources''': Paul/Ben/David
* '''QC''': ->
* '''Annotation & ontology''': no specific project
'''External'''
* CHEBI
* Complete proteomes
* Contributing groups


==Next hackathon==
==Next hackathon==
Line 68: Line 32:
https://docs.google.com/document/d/13BIc-zYaxIkIfK21Jhfk74pNq2jvtjepFdSGd6twpdw/edit#
https://docs.google.com/document/d/13BIc-zYaxIkIfK21Jhfk74pNq2jvtjepFdSGd6twpdw/edit#


=Minutes=
Chris slides - overview


Objectives:
*Need to get all annotations out of Protein2GO, have under GO Central control
**Only curatable annotations would be in Noctua, so not IBA IEA etc (Master of IBA is in PAINT DB)
***This doesn't mean they can't be used in models, can be brought in like any annotations currently can specifics need to be worked out (will this make duplicates?)


Idea: IEAs brought into Noctua converted into ISS since they've been manually reviewed- no, because ISS needs a w/f.  Use a new/different code? Generally agreed this is now a manual annotation (manually reviewed)
*Is a priority, but no dates/deadlines yet
*Difficulty due to variations in causal connectivity & scale


[[Category:GO Consortium]] [[Category:GO Managers Meetings ]]
Noctua database based on models: everything belongs to exactly one model
*this doesn't work with standalone annotations
 
# Refactor architecture so 'model' is optional
# work with this for now and hopefully improve in the future
Scenario: what if models are merged, then changes made upstream... what happens?
 
Would like to do experiments for different types of grouping (1 model/statement or gene up to 1/genome), but that takes a lot of resources and time
Problems likely with all options, so some experimentation is needed
 
;Proposal:
 
:1 model/gene is way to start
:Start on worm (maybe dicty?)
:Test case: can groups review/edit existing annotations (tool name currently Annotation Review)
 
How are annotations brought in, say, from Intact?
 
New models ought to be based on processes, not gene-by-gene, although seems likely curators will want to go gene-by-gene
 
Experiments on dicty need to be run on Noctua Dev
* Do some QC to make sure Noctua resembles intent of curators
* tool needs to handle large amounts of data (MGI beta catenin)
 
Rough timeline should be available by Cambridge, including working Dev version with a genome loaded
 
Why aren't AGR curators currently using Noctua?
*No ability to review
*presentation to users- goes back into GPAD, so time in Noctua isn't yet worth it
 
 
==Plan for moving legacy annotations to Noctua db==
* Strategy:
** start with (Paul proposes NOT to do dictybase, since it's not in Alliance) and wormbase, perhaps also MGI (to have some Alliance members)
** Move ALL Their annotations in (to discuss- IEA and IBA)
** David and Kimberly will select a set of genes to test whether the load worked correctly
* Timeline for the first database?
* Goal: Getting WormBase data in by the next GO meeting
 
==On call==
Chris, Seth, David, Kimberly, Paul, Suzi L, Suzi A, Judy, Ben, Jim, Huaiyu, Laurent-Philippe, Pascale
 
[[Category:GO Managers Meetings]]

Latest revision as of 10:32, 9 April 2019

Agenda

Annual report

What do we need from MODs ? We have been asking staff, personnel changes, publications, curation statistics from the current year. Looks like we have only included publications and overall number of publication curated in the last report. How about asking:

  • Number of FTEs on the project ?
  • Publications and presentations
  • Any highlights they would like to share in the annual report
    • Do we need any additional information ?

Follow up on priorities

  • Release: came out on Nov 16:
    • Suzi A, will you announce ? Do we have a process?
    • Why was it delayed? Do we need more resources on releases ?
    • Release tasks update (Seth)
  • GO website migration: Suzi L/Laurent-Philippe: Aim: Dec 1 -> Is this still the plan ?
  • Noctua+GO-CAM repository: Moving legacy annotations to GO central: Jim, Ben, Chris, Seth: strategy


Next hackathon

Early February in LA?

Proposed aims (David, Kimberly, Pascale):

  • import of legacy annotations & GP2term relations
    • also for importing legacy annotations - mapping existing annotation extensions to GO-CAM models

Action items

https://docs.google.com/document/d/13BIc-zYaxIkIfK21Jhfk74pNq2jvtjepFdSGd6twpdw/edit#

Minutes

Chris slides - overview

Objectives:

  • Need to get all annotations out of Protein2GO, have under GO Central control
    • Only curatable annotations would be in Noctua, so not IBA IEA etc (Master of IBA is in PAINT DB)
      • This doesn't mean they can't be used in models, can be brought in like any annotations currently can specifics need to be worked out (will this make duplicates?)

Idea: IEAs brought into Noctua converted into ISS since they've been manually reviewed- no, because ISS needs a w/f. Use a new/different code? Generally agreed this is now a manual annotation (manually reviewed)

  • Is a priority, but no dates/deadlines yet
  • Difficulty due to variations in causal connectivity & scale

Noctua database based on models: everything belongs to exactly one model

  • this doesn't work with standalone annotations
  1. Refactor architecture so 'model' is optional
  2. work with this for now and hopefully improve in the future

Scenario: what if models are merged, then changes made upstream... what happens?

Would like to do experiments for different types of grouping (1 model/statement or gene up to 1/genome), but that takes a lot of resources and time Problems likely with all options, so some experimentation is needed

Proposal
1 model/gene is way to start
Start on worm (maybe dicty?)
Test case: can groups review/edit existing annotations (tool name currently Annotation Review)

How are annotations brought in, say, from Intact?

New models ought to be based on processes, not gene-by-gene, although seems likely curators will want to go gene-by-gene

Experiments on dicty need to be run on Noctua Dev

  • Do some QC to make sure Noctua resembles intent of curators
  • tool needs to handle large amounts of data (MGI beta catenin)

Rough timeline should be available by Cambridge, including working Dev version with a genome loaded

Why aren't AGR curators currently using Noctua?

  • No ability to review
  • presentation to users- goes back into GPAD, so time in Noctua isn't yet worth it


Plan for moving legacy annotations to Noctua db

  • Strategy:
    • start with (Paul proposes NOT to do dictybase, since it's not in Alliance) and wormbase, perhaps also MGI (to have some Alliance members)
    • Move ALL Their annotations in (to discuss- IEA and IBA)
    • David and Kimberly will select a set of genes to test whether the load worked correctly
  • Timeline for the first database?
  • Goal: Getting WormBase data in by the next GO meeting

On call

Chris, Seth, David, Kimberly, Paul, Suzi L, Suzi A, Judy, Ben, Jim, Huaiyu, Laurent-Philippe, Pascale

Retrieved from "https://wiki.geneontology.org/index.php?title=Manager_Call_2018-11-21&oldid=74104"