LEGO October 17, 2016
From GO Wiki
- 1 Bluejeans URL - NOTE NEW MEETING TIME: 8:00am PST
- 2 Agenda
- 3 Minutes
- 4 Annotation
Bluejeans URL - NOTE NEW MEETING TIME: 8:00am PST
Monday Morning Model Disappearance
- We need to address the disappearing models on Monday mornings.
- Still working on this with Wednesday group
- Not all traditional GO annotations are being generated in the GAF/GPAD output
- Specifications are decided; need to update documentation and share it with rest of GOC
- Review specifications
- Groups will need to start submitting these files routinely, but there are still a few aspects of the specifications that need to be clarified and then documented on github, website
- Chris' gpad-gpi docs on github
- Kimberly's Google doc on gpi specs for website
- Contents of header - what's required, what's optional, format for tag value-pairs, format for free text
- !gpi-version: 1.2
- !Date: 2016-10-10
- Other possible tag value-pairs:
- !Project_name: WB
- !Contact_email: email@example.com
- !URL: http://www.wormbase.org
- !Release: WS256
- Other possible content:
- Explanation of gpi file fields
- Explanation of entities represented, e.g. Reference Proteome
- Explanation of specific DB_xrefs included
- Contents of file - what's required, what's optional, format
- Required fields are indicated in github doc, website
- Entities needed for curation should be included in gpi file
- Clarification needed on:
- Entity namespace in Column 1? - YES
- DB Object Type - use SO terms (and IDs?) here?
- Taxon prefix, i.e. 6239 or taxon:6239?
- Parent IDs - always a gene? canonical protein and a gene?
- DB_xrefs - are any DB_xrefs required?
- MODs should include xrefs from genes to UniProtKB GCRP (Gene-Centric Reference Proteome) accessions? Reciprocal?
- RNAcentral accessions for ncRNAs when possible
- Protein/macromolecular complexes?
- Gene Product Properties - anything required?
- UniProtKB includes information on Swiss_Prot vs TrEMBL subset
Use of PRO Identifiers for Modified Forms of Proteins
- David presented Noctua to the PRO team last week.
Domain, Range Constraints for Relations in Noctua
Tuesday Annotation Calls
- Next annotation call is a consistency exercise from RGD.
- Compile list of specific templates that would be useful to have.
- Use github projects to manage creation of templates.
- On call: David H, David OS, Giulia, Helen, Jim, Kimberly, Moni, Paul T., Sabrina, Seth, Stacia
Monday Morning Model Disappearance
- A result of a fail in the AmiGO load - SOLR gets flushed periodically
- Seth will manually fix this right now, but try to make Noctua model population independent of AmiGO build?
- Models have not gone away, they just don't show up on the main Noctua landing page
- Alternative way to get to models: open up Danger Zone link (this accesses minerva directly)
- This will be permanently fixed by the GOC meeting
- Still working on this on Wednesday calls
- Not all traditional GO annotations are being generated
- Only those annotations that are directly associated with an annoton are generated
- AI: Check go-help question about GAF/GPAD - essentially when will GPAD annotations become available?
- Specs are in place, just need to shore up documentation
Presentation to PRO
- David presented Noctua/LEGO to PRO
- Liked it but had two questions:
- Are all annotations in LEGO?
- How far across the property chains are annotations generated?
Importing Existing Biological Process and Cellular Component Annotations
- See Paul's github ticket - detailed specs on how this might work
- This is high priority
Domain and Range Constraints
- Will be incorporated as part of Noctua
- What will be the source for these?
- Chris has implemented a client-side java reasoner - this affects what properties are available to the user
- There are two templates to look at as examples
- Specify templates and design patterns at the same time would be a good use of efforts
- List of conserved processes that are good candidates for creating templates being put together by David and Kimberly
Presentation to AgBioData Consortium
- Will present Noctua to this group
- Discussion of model on Rian lncRNA effects on transcription
- Mechanism is not known, i.e. it's not known whether Rian activity is part_of transcription or regulates transcription
- Therefore, the relation between Rian and transcription is causally_upstream_of_or_within
- What is the effect on analysis if we do/do not include these kinds of annotations in GO?
- These Rian-type models may be helpful for network analysis, but for detailed analysis of biology, maybe not so much
- Should these annotations be included in the default annotation set?
- Should less direct annotations be included in the default set?
- Future iterations of AmiGO will allow users to more easily select different sets of annotations for their analyses
- Additional causal relations will be included in GPAD, could also be included in the Qualifier column of the GAF
- Another point about this model - the Rian molecular activity is connected to different transcriptional instances with multiple targets
- Will we always need to clone the molecular activity for each instance when there is more than one input for the transcription?
- What query use cases could address this question?