2018 Montreal GOC Meeting Agenda
Agenda moved to https://docs.google.com/document/d/16KMJRn-Vqf4yRsXAR1KYlwYvPob1tRTQbq2cEs6IJNY/edit#heading=h.uxqb7a8gufke
This below is out of date
Paul Project status, main accomplishments and discussion topics for the meeting
New Pipeline & new pipeline documentation
Kimberly Must be completed (as much as possible) and announced
- Report monthly PAINT updates and GAF file generation QC (Huaiyu/Dustin) https://github.com/orgs/geneontology/projects/23
- Plan for migrating to GPAD (Kimberly)
- Reports from each group
- Manage migration of consortium members to use explicit snapshot PURLs: https://github.com/geneontology/go-site/issues/783
- Excluding MOD-specific refs from the GO database: see https://github.com/geneontology/go-site/issues/791
- Fate of external2go mappings (Chris): https://github.com/geneontology/go-docs/pull/27#issuecomment-426311327
Differences between old releases and new releases
- Now releasing as a bundle
- Implications for various groups- including other distributors of GO data (QuickGO, NCBI)
- Improves (or just enables!) reproducibility
GO rules update and error reports
Eric & Pascale
- Documentation of rules: Which annotation validation rules are currently implemented? https://github.com/geneontology/go-annotation/issues/1928
- Current rules status & future priorities
- Status taxon checks (https://github.com/geneontology/go-site/issues/660 and https://github.com/geneontology/go-site/issues/758)
- New Error reports: http://snapshot.geneontology.org/reports/gorule-report.html (daily)
GO-CAM and Noctua
Noctua 1.1 release
- Updates on progress towards the 1.1 release
- Bug fixes
- New features
- Groundwork for future releases
Annotation review form
GO-CAM integration with other pathway resources
Roadmap for next Noctua release
- Integrating all (more) genomes in GOC
GO website migration
Laurent-Philippe/Suzi A: https://github.com/orgs/geneontology/projects/22
Ontology and Annotation documentation update
David, Kimberly and Pascale random thoughts:
- Ontology: mention the creation of 'projects' in GH where we moved old projects, so that its' easier to find old discussions
- future plans
Gene Product to terms relations
https://github.com/orgs/geneontology/projects/13 (Kimberley, Dustin)
- Resources for Annotation Review? Is reviewing a priority compared to creating new annotations?
- Signaling 2017 update
- ECM reviews
- Transcription reviews
- Transcription guidelines: https://docs.google.com/presentation/d/1hOBMnpHbGyKsOJi-Ah3xhAavzOg2RgM4AXRdGbF8A6M/edit#slide=id.p1
QC group report
- Add topics (report on progress)
How to blacklist papers
- List/form template of excluded papers or combinations of PMID/Gene/GO/comment, including retracted papers
- Centralized paper tracking system
- MF refactor including transcription: https://github.com/orgs/geneontology/projects/25
- ECM update
- Stop creating and start merging precomposed terms: x involved in y, X protein binding, protein complex x binding etc
- protein complex x binding: how does that impact 'colocalizes with protein complex x'?
- Pathways2GO: https://github.com/orgs/geneontology/projects/24
- Ontology maintenance: https://github.com/orgs/geneontology/projects/14
- All synapse tickets are closed
Binding and Protein complexes
- Should these be separate branches of GO ?
See (among other tickets) https://github.com/geneontology/go-annotation/issues/1940
- Proposal to convert all protein binding terms from IPI and a target in the 'with' field to IDA and a has_input.
GO subsets update
- Deprecated a number of unused/unmaintained subsets
- Show subset yaml files and how they are used
- Each subset needs a maintainer
Val: proposed new generic slim (15 minutes)
Defining redundant information
- Define redundant information:
In AmiGO, we should be able to improve the display by removing redundant information. That information may be useful for certain purposes, so we should provide it in files. We could also provide the 'core set' ('stringent set') in some version of files. There are tickets that explains the strategy that will be taken for this: see https://github.com/geneontology/amigo/issues/43 and https://github.com/geneontology/amigo/issues/440
- AI: Rules for flagging redundant annotations need to be documented - Chris
Use cases for 'redundant', non-experimental annotation
- Are there any use cases where people have used these annotations for some type of analysis?
- Some pipelines (InterPro2GO, SPKW, PAINT, F-P links), sometimes provide data that is already captured experimentally, and some groups would like the redundancy reduced.
- Should all GOC members be handling redundancy in the same way?
- If redundant, non-experimental annotations are present and are going to be removed, at what point in the pipeline should they be filtered, e.g. annotation file production by GOC, annotation file processing by MODs, website display?
- If we filter annotations files, should we then also provide two annotation files for users, one complete and one filtered?
- Doubled up IBA+EXP annotations (from Karen Christie)
- Issue with GOC inference file (i) incorrect aspect reported
What data to present to which users
Examples (different types of redundancy):
- TAS http://amigo.geneontology.org/amigo/gene_product/UniProtKB:P05177 (13 annotations to the same activity and 10 to the same component by TAS)
- Exact duplicates by GOC function-process pipeline
- Identical annotation by different groups (rare)
Representing complete proteomes in GO (added by Val, as suggested by Chris)
Datasets (obtaining/maintaining complete datasets with unique identifiers)
Overview. Nobody seems really sure what happens. I'll document what I think happens here and then run it by others to confirm
1. GO uses https://www.ebi.ac.uk/reference_proteomes to define the set of human IDs uniquely. This is also used by Panther. This reference proteome set represents each HGNC ID uniquely. This causes issues when 2 proteins are encoded by the a loci described by a single HGNC name.
2. UniProt has other versions of reference proteome (I asked UniProt helpdesk about this).
3. What to do about READThroughs from Ensembl (MED17 as an example)
How do changes in the reference proteome in-between releases affect GO? i.e What happens to new or revised IDs if they are used in GO annotations, but are not represented in the reference proteome?
~https://github.com/geneontology/helpdesk/issues/161~ closed moved to https://github.com/geneontology/go-site/issues/819 (this ticket is just a general question about which dataset we use)
- What does reference proteome mean, why we have duplicates for pufferfish, and why it includes unreviewed proteins? See https://docs.google.com/document/d/1l4hZw_ZTor8RoTjYTNS4FtGAba4HZrewmjccrSXJ9Hs/edit
- Which organisms other than cerevisiae and pombe have looked at all protein coding genes for the availability/possibility of GO annotation? Establish the difference between:
- (1) "not in the GO database (not found);
- (2) "unknown" (ND),
- (3) "unannotated" (no ND, and no annotation in Aspect of interest)
(difference can be established using the complete known protein ID set for your organism and GO term mapper https://go.princeton.edu/cgi-bin/GOTermMapper)
ChEBI ontology (as suggested by Gareth Owen)
1) Would the GOC be interested in simplifying the ChEBI Ontology?
Over the last year or so, we have been working on allowing individual users to 'personalise' the ChEBI Ontology, by choosing terms in the ontology that they don't wish to see (in effect, reducing the number of levels between the name of a molecule at one end and "chemical entity" at the other). This has to be done term by term, but we have also been working with the Ontology Lookup Service to enable this type of process to be automated.
2) There are a few GO-submitted issues in the ChEBI issue tracker on GitHub that could benefit from discussion (though I suspect that a discussion of a few people during a break may be more useful than involving everybody ia a separate part of the agenda?)
3) From a ChEBI point of view, a lot of effort goes in to building and maintaining the ChEBI ontology, but we would like to understand how the ChEBI ontology is then actually used (perhaps as part of GO) to tackle real problems. While we know that a number of ontologies (including GO!) make use of lots of ChEBI terms, we would be very interested in how the ontologies are actually being used in tools, applications, etc. in the 'real' world.
4) Give a broad perspective of ChEBI, how it links to other resources e.g. via UniChem and key applications e.g. in metabolomics.
Breakout sessions topics
- Specifications for next version of GPAD/GPI
- Guidelines for submitting annotations to GO - for example Ivan Erill also had an idea to ask the organizers of the Phage Meeting to provide an option for abstract submissions to include author-generated GO annotations. What would our guidelines be ?
- GO Slims - review Alliance slim with latest stats from Mary Dolan. Does the goslim_agr need any updates?
- Modeling transcription in GO-CAMs
- 'Response to' workshop (similar to the signaling WS)
- Use cases: Should we add this to the agenda? What would be a productive way of discussing this topic? https://docs.google.com/document/d/104m4jUNjPH9pCpskg8E29Zm2pLHhPFmKyIFLa9l_EOQ/edit
Product owners/tech leads discussion - lunch Thursday
- Small debrief session: Lessons learned, suggestions for improvements, next face-2-face meeting