Ontology meeting 2012-12-20

Minutes: Tanya

Attendees: Heiko, Paola, David, Tanya, Harold, Chris, Jane, Judy

Harold has issues

GO id is primary id for one term and a secondary id for another term. Came up in MGI load error message. MGI loads gene_ontology_edit.obo (not the editors' file). Recommend that they switch to loading from URLs (Chris will send details for purl) for future, though this is not likely the problem here. GO:0032779 is a primary ID and it is also used as an alternate ID for GO:0006886.

  AI: Tanya will look into this.  FIXED!

Renaming go_xp_chebi

It would be easier if we had a single file for all production-level logical definitions. We are ready to go with x-transport-and-localization.obo, it would be easiest to copy the stanzas directly in (and then run the obo normalizer).

We can create different output products for users - e.g. a file that is just intersection stanzas that reference chemical entities (can be done in owltools + Jenkins). We should not be pointing users at any files in the editors directory anyway.

If accepted we can do this on Thursday in Berkeley with T&D (will require momentary TG downtime).

  AI: Sound good. Do this in Berkeley.
  AI: need to change the link in the paper to match the new URL - currently references x-chemical.owl.
  AI: Add cation/ion information into ChEBI paper?

Wrap-up: ChEBI paper

Chris and David to report on action items from last week

(Chris to prepare flow diagrams; David to draft cover letter.)

  T & D will be enforcers.
  David wrote cover letter.

Wrap-up: Discuss default relationships for the qualifier column in the new GPAD format

Copying over last week's discussion/questions/to-do:

part_of------ Do complexes create a problem with this? Much discussion. If a complex is defined by its members is it always a part_of or is it sometimes attached to?

is_active_participant_in---- does this create problems when we have mutant phenotypes? This is reminiscent of the regulates conundrum where we annotate to the parent. Do we need a weaker relationship for process? It would be the union of is_active_participant_in and regulates_active participant_in. Should we use involved_in? We need to decide how annotators will choose these relationships.

  Yes, Rachael agrees with the weaker relationship.  What to call this?

  Chris is trying to figure out what this relationship means.  We will discuss more later today at Berkeley.

Follow-up: Mapping proton/hydrogen ion terms to the correct ChEBI term

Harold to report.

For background: copying over last week's discussion:

Stemming from a TG request for 'proton binding' (mapping to CHEBI proton). Based on the reference, the mapping should go to CHEBI oxonium (H3O+) instead.

Same for xxx:proton/hydrogen symporters/antiporters.

Is it always the case that whenever we are talking about a proton we are talking about a proton? Harold thinks that you can pass a proton itself. Harold will investigate. Are we going to have to lump the proton and oxonium together like we did for the acids and bases?

  REPORT:  According to expert, it is indeed the 'naked' proton that gets transported and not the oxonium ion.  
  When swimming around it is the oxonium ion, when transported, it is the proton.  We were needlessly worrying, 
  we can leave it as proton.

  AI: Harold to update proton renaming SF item (currently assigned to Paola) with this information. DONE.

Follow-up: TermGenie free form

Chris and Heiko to report on action item from last week:

(copying over) explore feasibility of preventing submitters from creating regulation terms if the target term is a descendant of 'regulation of biological quality'; Chris and Heiko will check the existing constraint to see why this doesn't violate 'regulation of regulation'. If we create a cross product for R of BQ then we should fit the constraint.

  How to create 'black list' of terms to add? Through the logical parentage?  Both for free form and templated targets.
  For the templated terms, perhaps we can use restrictions through the parentage of the target. Education of term creators
  is necessary.

  Suggestion: Have a point person per database/group who is responsible for TG addition for that group and trained 
  properly.  Similar to point person for CVS access.  Email to GO group asking for nominations for people to have 
  this responsibility.

Report on testing/advertising.

  REPORT: Additional testing largely successful. Inform select additional curators of this functionality. Roll out 
  after holidays.

Follow-up: Protein/glycoprotein/lipoprotein terms in ChEBI

Any updates on this, from either ChEBI or PRO?

  REPORT: Email reminder to Darren about PRO's plans.  ChEBI may move conjugated protein out from is_a protein and be
  has_functional_parent protein instead. Some issues may remain.

Documenting decisions

  How do we do this?  How about a GO:ref and then we put the URL to the wiki page with more info in the GO:ref?  Some 
  more general information could go into the ontology header.  More specific data can go into the stanza/s for the 
  relevant terms. Ask Tony if text from GO refs can be displayed in QuickGO?

  GO ref based solution is a good one.  Flesh out more.

  Berkeley discussion:  We will create GO_REFs liberally when necessary and use them as dbxrefs for the relevant parent
  terms. State in the GO_REF that it applies to the children terms.  May want to apply a new relationship 
  applies_to_descendant_terms.  For example, if decisions are made during content meetings, these can be documented in 
  a GO_REF.  This will help us in keeping track of why decisions were made and allow us to refer back to them in the future.

Wrap-up: Alerting users of long-term changes to the ontologies

(Copying over from last week) See background from last week: http://wiki.geneontology.org/index.php/Ontology_meeting_2012-12-06#Alerting_users_of_long-term_changes_to_the_ontologies

So can we label terms for specific projects, like the apoptosis project etc? Should we just make subsets?

  AI: Yes, we will use subsets to label these terms.

obo-diff.html

chris + heiko to report

  REPORT: on their way.

Submission(s) to the first annual JBMS Ontology Issue

go.owl

How often does it update? Nightly (midnight Pacific) if there is status is blue with the write file. If red ball then go.owl will not build. Can also do this on your own using obo to owl converter.

Redundant link removal

Jane will do these for the write version at some point.

cell cycle meeting

Jane/Val, no real date yet. Two dates in play, perhaps later. Cast of characters necessary. List will be created, inform Judy. Judy has suggestions as well. Some funds available for ontology development workshop. Possibly some funds for small obo to owl training for developers. Think about options before March 1st.