Meeting Minutes 2
Implementation Meeting 7th Dec, 2007
David Hill, Tanya Berardini, Chris Mungall
Went over results of work with Chris. Explained additional requirements for rerunning of analysis.
- Filter out the ROMF and RoBQ terms.
- Don't report on terms that we say are ok based on the discussion that follows.
Chris says the analysis is set up to run nightly but will add additional requirements as specified.
Question: Create terms in the function and process ontologies for which there are none now?
Example 1: 'regulation of ARF GTPase activity' exists but there is no 'ARF GTPase activity' in the function ontology. There is 'GTPase activity.'
Example 2: 'regulation of transcription from RNA polymerase II promoter involved in forebrain neuron fate commitment' exists but there is no 'transcription from RNA polymerase II promoter involved in forebrain neuron fate commitment.' There is 'transcription from RNA polymerase II promoter.'
Solution 1: Create the new terms.
Pros: Good for making the generation of logical definitions easy and straightforward. Cons: The underlying processes and functions of the two terms (existing and the proposed new child) are not different. ARF GTPase activity is no different from GTPase activity (the same chemical reaction is catalyzed). Same goes for the transcription term.
Solution 2: Don't make the new terms. The 'regulates' link will be realized through the parent term, rather than through the term itself.
Pros: Doesn't introduce unnecessary new terms into the ontology. Cons: Logical definition generation becomes more complicated.
Question: Should we start creating interontology links, specifically between process and function in GO?
Answer: Yes, but keep links in a separate file that can be loaded up by the editors. Do not make this available to the general public just yet.
This can be part of moving towards a multiple file system for the GO, with ones specifically for editing and one for public consumption.
Between now and January:
Chris: make modifications to report generation script so that additional requirements are taken into account
David and Tanya: examine resulting data and update ontology accordingly
Chris and David: at next managers meeting, bring up the 'pre-warning on regulates relationship' email to be sent to the community,bring up possible problems (icon changes, calculating closure over new relationship type, not any worse than current situation)
Chris, David, Tanya: make regulates real, realize the implied links, could do with OBOEdit2 or with OBOEdit1 (in two cycles) depending on state of OE2 at that point in time
attacking logical issues
Chris: generate report with missing links (includes those where the process DAG does not agree with regulation DAG)
David/Tanya: address structural issues, both of incompleteness and outright errors.
- links between GO and PATO (after regulates relationship is live)
- find RoBQ terms that don't have corresponding terms in PATO and request them through their tracker