Ontology meeting 2014-07-17

From GO Wiki
Jump to: navigation, search

Attendees: Jane, DavidOS, DavidH, Harold, Judy, Tanya, Heiko, Chris

Minutes: DavidOS

Alpha-glucosidase definition is more specific than its name

SF10984

  Agreed to genericise the term.

isa vs partof in GO and Uberon

we have cases where we need to resolve if isa or partof is the correct relation to use:

uberon has been modified to follow GO's classification except for a few cases where GO needs to change:

  David OS: We should generally defer to the AOs.
  Chris: But they often contradict each other
  David H: And where we've agree something in an expert workshop - maybe the 
  In general discussion all agreed: systems should be organism or organ-wide with sub-systems as parts rather than subclasses
  
 Chris to add notes on detailed decisions?

phases and cycles

I've added 'biological phase ; GO:0044848' as discussed last week.

I think we should use happpens_during between phases and cycles, and between phases and processes that occur during them e.g.:

  • cell cycle phase happens_during cell cycle
  • uterine wall breakdown happens_during menstruation

I also think for consistency phases shouldn't have regulation terms - i.e. you can regulate the processes that happen during a phase, but not the phase itself. The cell cycle phase regulation terms were already obsoleting but we have some regulation of hair cycle phase terms remaining.


   plan to use happens_during in both directions (phase -> process; process -> phase)
  Agreed
   regulation of phases:

get rid of these for cell cycle phases But for the others - change to 'regulation of timing'

we agree that hair growth is developmental


Everyone happy?

(as an aside, I need to add 'hair growth' - would you say this is a developmental growth?)

   Yes

Protege 5 testing

https://www.ebi.ac.uk/panda/jira/browse/GO-320

New search tool problems: Submit bug & feature requests or roll our own?

  AI: David to draw up feature request and submit as Protege 5 GitHub issue following input from GO editors. 
  AI: David Characterise GCI display in description. - which ones show up?

Annotating protein complexes - blurring the line between the ontology and annotation

Rachael asks

   "for annotation of a protein complex entity to a GO:protein complex term, the default annotation 
   part_of doesn't seem correct as you would be saying the complex is part_of GO:<x> complex. 
   Should we be using is_a in these cases?"

I replied:

   "There could be an is_a 'relationship' between an annotated protein complex and one in GO, 
    but this is not guaranteed, so would need to be recorded by the annotator (probably as a qualifier)."

I had a meeting yesterday with Sylvie Ricard and Birgit Meldal on terms for ECM components & things that bind them. There was much confusion about what counted as annotation and what goes into the ontology. Not surprising when we have complex term for homotrimers of collagens, and these have part relations to other, larger complexes and have their function recorded via capable_of / capable_of_part_of relations to MF / BP. i.e., in these cases we already have something like CC, MF & BP annotation for collagen gene products (at least when they are part of homotrimers - which is much of the time.

Both of these examples are symptoms of a blurring of the lines between annotation and ontology building. I think we need a longer term plan to manage the division of labour here. Should we be aiming for protein complexes to become the responsibility of Intact + annotators rather than being in GO?

   Poss meeting that brings together Pro/Intact people with GO to decide how to divide up the work and make it sustainable.
  Chris: If you want the generic protein complex in GO it has to either 
  (a) be functionally defined or (b) defined by its parts  (or both?)
  but if it is a part defined complex there must be one at least one species specific one in either Pro or Intact. 
  We would have an axioms that points to one or more intact classes (as prototypes?)
  David: But how does this => us useful inference?
   Alternative - specify some minimal def of what must be part of a complex to fit generic multi-species class. 
   Allows for additional components
   Needs to be able to reference domains or protein families.

Plan for annotation extensions relations file

Move to OWL with RO imports?

  Chris to make OWL version with import mechanism for RO, retaining annotator definitions
   as comment or some other Annotation property value.