Ontology meeting 2012-08-22

From GO Wiki
Revision as of 14:28, 1 July 2014 by Gail (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

45-minute meeting


ATTENDEES: Becky, Jane, Judy, Heiko, Chris, Harold, Tanya, David

SF triage and TG Gatekeeper Rotas

Extended rotas are at:

Asserting implied links/removing redundant links for chemicals

  • For the work with CHEBI we really need to assert all of the links implied by go_xp_chebi.obo. Would also be good to remove the redundant links while we're at it. Can we come up with a plan of action to do this asap?
    • Note that these are added/removed in the release version of GO anyway
      • Option 1 (Xtreme):
        • Ditch Obo-Edit for all editing. Use Protege with Elk, always viewing the inferred hierarchy
      • Option 2:
        • add asserted links and remove redundant links in editors version (not a good long term strategy)
        • possible to do this with OE, but extremely kludgy
        • best option: take Oort output (i.e. go.obo) and copy it back into editors - TO BE DONE BY BECKY
        • may want to do this more than once
        • right now can't tag relationships that are not supported but we want to retain, to distinguish these from unsupported relationships that need to be removed
        • future: can tag relationships - not in OE but under the hood, would need to be filtered


  • Carried over from last week's meeting [1]. What's the consensus on having roles in GO?
    • morphogen activity ; GO:0016015 (signaling activity involved in morphogenetic process as the logical definition)
      • should be classified automatically. E.g. activates some signal transduction and part of morphogenesis
      • anyone querying for morphogens in GO right now will get an under-representative set
    • hormone activity
    • protein tag ; GO:0031386
    • holin activity ; GO:0034290
  • If roles are well-characterized and they have a part_of relationship to some process, the term can be created.
  • If mechanism is not known, do not create a function term.
  • Separated information: Entity X has role R in species S.
  • David will write proposal for how to create terms to distinguish roles that are not universal.

New template for 'protein localization to'

  • Carried over from last week's meeting [2].
  • We have a mixture at the moment of 'localization TO, localization AT, localization IN'. We should standardize these, so what's the best wording, and what do we mean by it... presumably a protein travelling TO an organelle/cellular location and travelling/being maintained IN that organelle/cellular location once it's there.
  • Definitions are more or less the same.
  • Becky will standardize all term names to "to" rather than "in" or "at" x location. DONE:23.08.12.
  • Chris' SOP looks good to implement.
  • next up: response to x chemical, correctly parsed xps are now in the go_xp_chebi.obo file

Term Genie-related issue

  • Molybdenum co-factor term that was put directly under metabolic process instead of co-factor metabolic process because TG does not use roles.
  • Solution: have logical defs for upper level 'role' terms
  • For now, commit TG placement, then edit in OE and make a comment that covers why changes were made. As a good audit trail.
  • Other option was to fix relationship as TG gatekeeper before commit but we lose the information that TG put it too high in the graph if this was done.

single-organism process v/s multi-organism process

As discussed at the last meeting, we've been overloading the meaning of 'cellular process' to mean both 'a process occurring at the cellular level' and 'a process occurring within a single organism' by putting a disjoint between cellular process and multi-organism process. This causes problems for cellular-level processes that involve more than one organism, as we've found repeatedly with viruses. Becky and I propose the following top-level split:

multi-organism process
---[i] multi-organism cellular process
------[i] interaction between cellular organisms
single-organism process
---[i] cellular process within a single organism 
---[i] process within single multicellular organism
cellular process
---[i] cellular process within a single organism 
---[i] multi-organism cellular process
multicellular organism process
---[i] process within single multicellular organism
---[i] interaction between multicellular organisms

Then the disjoints would be between cellular process/multicellular organism process and multi-organism process/single-organism process. We could probably find better names.

GO for it, Jane and Becky!

suggested logical def for cellular process

intersection of: biological process, GO:xxx intersection of: has_participant cell CL:xxx

ChEBI paper update

Chris has read through it again. Will have edits ready for tomorrow, meeting w/ D/T.

Multi-functional enzymes

For example, terminase has two functions: translocating the nucleic acid into the phage head, and cleaving it at the correct point. These are always coupled.

We get these issues in SF all the time, so we need to come up with a general solution, even if it's not likely to be our final solution.

A related issue is where the same or similar reactions can be carried out by one enzyme v/s multiple enzymes. We've handled this inconsistently in the ontologies - sometimes we have one term for the overall reaction (especially where EC do this e.g. methylthiol:coenzyme M methyltransferase activity ; GO:0044683), sometimes we break it into its constituent parts.

  • Difference between what the enzyme is doing and the individual functions that it performs.
  • Possibly consistent solution: one compound function term with has_part children for the individual functions, hard part might be coming up with the name
  terminase activity
  --[has part] nucleic acid translocating activity
  --[has part] nuclease activity 
  • If both (all) are required and have to happen in concert for that function to occur, then it can be considered a real function and worthy of adding to GO MF.