Ontology meeting 2014-04-24

From GO Wiki
Jump to: navigation, search

Attendees: Jane, Harold, DavidH, Paola, Chris, Heiko, Tanya

Minutes: Tanya

Extracellular vesicle terms

See email thread "GO and vesicle nomenclature"

Report on discussion from yesterday's SF call:

 TODO: Jane to contact group working on these to ask for clarification 
 about two intermediate classes which are both parent classes to all leaf nodes in 
 membrane bound extracellular vesicle branch.

Relations files

Talking with Rachael today, there's some confusion about the various files that contain relationships (go_annotation_extension_relations.obo, gorel.obo/.owl, RO) and how they relate to one another. It might make sense to have a subset of go_annotation_extension_relations.obo that would be available in protein2go for use in annotation extensions, because there is a lot of noise in this file.

  Extension relations need to move into RO.  go_annotation_extension_relations.obo is manually kept in synch with RO, has extra 
  stuff in it about domains and ranges.  Need to create a subset that people will want to use for annotations extensions, chains can 
  be hidden.

Follow-up: 'colocalizes with' nuclear chromatin MOVE TO NEXT WEEK

Carried over, see http://wiki.geneontology.org/index.php?title=Ontology_meeting_2014-03-13

Where are we with the planned action items? David OS made a start on this. Transitive?

The latest AI was for David OS to put together a proposal for this, and run it by Tony.

Follow-up on Protege

Are we really ready for Protege? Copied over from previous meetings:

David OS will add examples and tips for GO editing and browsing in Protege on the wiki. Chris has started asking people about the plug-in. The plan is to have a meeting to figure out what kind of progress we can have on that in a reasonable time-frame. We need a write-up of the sort of things that are currently hard for editors to do/render in Protege; things that you can do in OE but can't currently do (or easily do) in Protege. E.g. new terms, searches... rendering in tree views is virtually impossible now. We could have a period where we do things using both tools, but one worry is that some of us may not be able to work with both tools open on a machine that doesn't have loads of memory.

  AI for all: WISH LIST For the time being, we can all try using Protege in our normal work flow and report 
  on the difficulties we encounter. Jane made a wiki page and we can all add to it: 

AI: DOS to add link to the specs document.

Specs doc is here: https://docs.google.com/document/d/17tph3jwrDGEaa7DyYPOK8HCi2dCDdJxi3VrTWNQpbqc/edit

Alpha version of OBO annotation plugin is here: https://github.com/simonjupp/protege-obo-plugins

The wiki page will be the place to collect use cases, the Google doc will be for specific feature specifications.

Any progress on this?

  • Update from Chris on "Plugins that ease the transition from Obo-Edit to Protege" (sent to the OE mailing list on Apr. 21st):

The GO has been maintaining a list of Protege plugins that are of interest to the GO group as we gradually transition to a more OWL oriented workflow, many of you are in the same boat and would find these of use:


Thanks to Jim Balhoff, Simon Jupp and Heiko Dietze, the core developers, and to David OS for pushing us along.

Many of you know how to install plugins in Protege; for everyone else, we're hoping to aggregate some of our existing training material into something easily digested, so stay tuned, and don't be discouraged if you're not sure where to start.

Note that in the list we have an alpha version of a new plugin that makes it easier to edit OBO-style metadata: synonyms, definitions and comments. I know many of you like being on the cutting edge, so we're letting you see a sneak peek (I'm already finding it useful), but we're not asking for feedback on this plugin just yet, as we know what the issues are and will be addressing them in updates over the next month.

Follow-up: Reactome ontology mappings

Carried over from April 3rd: Reactome maintain mappings between their processes, reaction etc and GO terms. We have mappings to Reactome in GO, but these are not in sync - one of Amelia's legacies I think. How do we want to handle this? The nuclear option would be just to remove the mappings from the ontology file, or alternatively we could set up a regular job to get the GO-Reactome mappings from their BioPax. Thoughts?

Heiko and Chris will set up a job to make these automatic. Jane will open a Jira ticket.

Where are we with this?

  Need a way to keep these in sync.  Suggest using human or generic mapping.  Peter may have suggested not using the generic one -
  Chris will check with him.

Demo of Protege plugin for editing terms

Made by Heiko. Makes it possible to edit terms in Protege with a bit more ease.

Detecting terms with no definitions

Need ticket for Jenkins to check for terms with NO definitions.


Mapping for company X

  Continuing conversation with Jane.  More work, will get some money for the work.  Will likely create union classes to match their categories, 
  for example, adipose cell process.
  Use owl for maintaining the file - equivalence class to named class or arbitrary union or intersection of terms
  They want a simple text file.  They want to be able to edit the file - this would require editing the owl.  May need to train them and
  give them access to svn.  WebProtege may work for this.  File will be public.

Rest of JIRA issues

  AI: Heiko Need to turn back on the inferring of links for ontology.  Don't remove redundant links for now until we've resolved
  whatever the bug is that's messing up the ETINEs.

Try to review those we didn't get to last time.

  Review done, updates made directly to the JIRA issues, not captured in these minutes.
  • Update from Chris (Apr. 21st):

I'm currently trying to group the various tasks we're working on into grouping tickets, such as:

https://www.ebi.ac.uk/panda/jira/browse/GO-222 (Migrate editors file to OWL)

https://www.ebi.ac.uk/panda/jira/browse/GO-217 (Taxon constraints)

https://www.ebi.ac.uk/panda/jira/browse/GO-255 (LEGO environment)

https://www.ebi.ac.uk/panda/jira/browse/GO-190 (RHEA)

These are not all necessarily the purview of the ontology group - e.g. LEGO is a pan-GO project, but the Ont Group are crucial here, e.g. defining relations, adding starts-with and ends-with axioms.

Maybe we should get more into the habit of using these to structure some of our calls? I feel that sometimes some things slip through the cracks as we have so much on.