Ontology meeting 2011-10-26

From GO Wiki
Jump to: navigation, search


Discussion notes

Protege/OWL training session

  • We need to find a place and time soon. The meeting has to be before the end of February.

GOAL: at the end of session, editors should be working in dual mode, know benefits of both tools, know which tool is the best for which job. Similar to how David OS works on FlyAnatomy - very fast, sequential editing - edit in OE, reason in Protege. If we do this, we can have more power/knowledge to request features or plugins for Protege.

Good for DavidOS to attend? If yes, then training location might be better in Cambridge/EBI.

Judy: spend funds before end of February (will confirm), annotation camp at Stanford in March likely (buy tickets before end of Feb.), so maybe have

Chris: no last week of Feb/first week of March

Possible time: [week of Feb. 6, week of Feb. 13] January 13-17, 27-31 (fly in Sunday, meet Monday/Tuesday, fly out late Tues/early Wed) Jane will look at room availability at EBI. Two day meeting.

Theoretical/training on Day 1, practical ontology development on Day 2. Pick something that we can write good logical definitions.

TermGenie 1 Update

  • Chris: demo at GO meeting, save some current requests for testing TG1

Discussion notes I


  • Dianna has submitted a bunch of SF requests for 'regulation of transcription in response to x' terms.

E.g. Reg of txn in response to acidity

  • They're currently all assigned to David. Please can we have a new TG template for these.
  • Looking into this further, existing terms of this syntax currently have 2 is_a relationships to the parents. Is this correct.... isn't transcription PART_OF the response?

Should be PART_OF, therefore 'involved in' instead of 'in'. TG template should be very similar to the 'involved in' one. Primary term name will contain 'in', synonym will be 'involved in'. If 'response to x' doesn't exist, create that term. All of them will be linked to 'cellular response to x' but term phrasing will exclude the 'cellular' word but use the 'cellular' definition.

Term: A involved in response to x

Rule: If term A is a cellular process then the response to x should be a cellular response to x.

Possible solution: select cellular term then hand edit the term

response to stimulus ; GO:0050896
Any process that results in a change in state or activity of a cell or an organism (in terms of movement, secretion, enzyme production, gene expression, etc.) as a result of a stimulus. The process begins with   detection of the stimulus and ends with a change in state or activity or the cell or organism.

Also: (revisiting our favourite topic from previous meetings):


Since we haven't been able to find a good solution for the XPs for these, could we have a template in TG to add these in with two is_a relations, and XPs to match the existing terms. Then we can fix them en-masse later, when we can capture primary regulation, secondary regulation, tertiary regulation etc ? It's a fudge, but would mean the annotators get an ID for their terms. There are loads of these in SF which would be very nice to close.

See positive regulation of gluconeogenesis by negative regulation of transcription from RNA polymerase II promoter ; GO:0035949, for an existing example.

Discussion notes II


  • (Carrying this topic from previous agenda, as we'd like to hear Chris' feedback on it)

A plea for a re-cap on what information is needed to make a cross-product, in reference to the involved_in terms.

See http://sourceforge.net/tracker/?func=detail&aid=3396697&group_id=36855&atid=440764

Editors were choosing most specific parents for the logical definitions that we could. Ontology is not exhaustive so choosing grandparent term for logical definition might result in terms with the same logical definition.

Better to have single genus and single differentia. Some definitions may need two differentia.

Discussion notes III

Coming back to an earlier discussion: How are the 'regulation of' terms related, when the two parent processes are connected by a part_of relationship?

Clarification required for: https://sourceforge.net/tracker/?func=detail&aid=3317785&group_id=36855&atid=440764

  • Becky's closing response on the tracker sums this up. Note that the current behavior derives from the transitive_over assignment in the GO relations, i.e. regulates o part_of --> regulates. Cjm 13:46, 25 October 2011 (PDT)

When requesting a term 'regulation of pigment cell development', TG does not create a relationship between 'regulation of pigment cell development' and 'regulation of pigment cell differentiation'. Surely it should?

Or, the reasoner should be able to do it. Chris, is this the case?

  • It should. I can guarantee that TG1 will provide a complete classification, as it uses 3rd party reasoners that have been verified using mathematical proofs etc. The TG0 classification should always be correct, but may be incomplete for the transitive_over cases. Apologies for this. These can be classified more specifically in OE using the RBR, which is guaranteed to use transitive_over. Also when we switch to using Oort for the public releases, the public release is guaranteed complete. Cjm 13:49, 25 October 2011 (PDT)

Task List