Ontology meeting 2012-10-18
ATTENDEES: Jane, Paola, David, Harold, Heiko, Tanya, Judy, Chris
FOLLOW-UP: ChEBI paper
Did the PIs give us any comments?
David received comments from Paul S. and Paul T. He passed them on to Tanya who's next.
Want to submit paper by end of next week - PIs have been notified. They'll send comments by end of this week and Tanya will collate them.
Timing of tracker call
For Paola, Jane, Tanya and David to discuss troublesome tracker issues.
Wednesdays 8:30 - 9:15 on weeks when there is no GO managers' meeting.
Thursdays on weeks when there is a GO managers' meeting, incorporate into GO editors meeting.
Since there was a GO managers' meeting this week, we'll have a SF pow-wow next Wednesday (10/24).
FOLLOW-UP: import/export template
Where are we on this? Chris, have you defined the relations we need?
On a related, good-practice-rule note: we will never use 'uptake' in GO names if it is within an organelle.
Discussion item: cell-cell junctions
Related to . Basically, do cell-cell junctions have parts that belong to each of the cells they bridge, or are they part of just one of the cells?
Right now we have defined and represented them as one cell. See the definition of cell junction and the placement of terms like cell-substrate cell junction.
If we say 'cell junction' has_part 'plasma membrane', is that logically the same as saying 'cell junction' has_part SOME 'plasma membrane', or 'cell junction' has_part WHOLE 'plasma membrane'? Would it better to say 'cell junction' has_part 'plasma membrane part', or is that logically equivalent to saying 'cell junction' has_part 'plasma membrane'?
Suggestions: move 'cell junction' directly under cellular component; create 'cell junction complex' under 'plasma membrane part'; create 'plasma membrane part of cell junction'.
AI: David will fix this.
Discussion item: non-ChEBI parents for chemical terms
Related to  chemical triad + regulation terms added, but no secondary metabolite M/B/C parentage
We need to look through roles, and figure out which ones are ok to always have as parents, and which ones aren't. The consensus today is that 'drug' is more difficult, but a 'secondary metabolite' would usually be true for all species.
AI: Chris will generate a list of roles that have a match in GO; this might be missing something, but we'll realize if so.
AI: Jane to look at that list.
Did Heiko and Chris make any progress getting Stanford to change whatever they had to change so we can move forward with this?
FOLLOW-UP: logically defining single-/multi-organism processes
We discussed last meeting defining:
cellular process = biological process has_participant exactly 1 or 2 cells multicellular organism process = biological process has_participant 2 or more cells
This won't work for cases like flocculation, which is a cellular process which involves >2 cells.
Could we say:
multicellular organism process = biological process occurs_in multicellular organism and has_participant >2 cell
but then for cp:
cellular process = biological process has_participant exactly 1 cell OR has_participant >1 cell AND occurs_in ???
I think this is the problem - how to distinguish between cellular processes that occur within a multicellular organism, and multicellular organism processes. Perhaps use tissue? e.g.
cellular process = [biological process has_participant exactly 1 cell] OR [has_participant >1 cell AND occurs_in 1 tissue type within a multicellular organism] OR [has_participant >1 cell AND external to a multicellular organism]
Can you express that in OWL?!
Sorry, that's a ramble of an agenda item.
Discussion: David thinks that this might not work because of cell signaling sometimes occurring across tissues.
We all agree that the cell is the major player in the process, even if things from the outside can come in and influence that process.
Could do has_participant some particular cell types?
Or we could do 'cell-cell signaling' NOT a cellular process; same for 'cell junction organization'. We'll have to run this past Becky. And we'll try to incorporate the tissue distinction.
Chris says that to express that in OWL we'd need to resort to 'tissue portions'.
We'll return to this.
AI: Jane to speak with other people to see if they have more suggestions.
Discussion item: MOP/CP - behaviour
An item related to the above. We think 'response to...' processes are always single-organism. This is because the process starts when the stimulus has been detected. It doesn't include the stimulus, even when that stimulus is another organism.
At the moment, we have behaviour classified as a response to stimulus. However, there are some behaviours that are definitely multi-organism. For example mating - this inherently has two participants!
So I think behaviour needs to be moved out from response to stimulus so I can split it into single and multi-organism behaviours. What do you think?
FOLLOW-UP: TermGenie template for non-template-able terms
See background here: http://wiki.geneontology.org/index.php/Ontology_meeting_2012-10-11
Question: Taxon triggers
Becky was doing it, but we should all be doing it more prospectively...
- to add a new one, first request an ID range for tax rules
- to remove, comment out (!) for now - later we will put in
FOLLOW-UP: EC numbers in GO - feasibility study
Carried over from a previous meeting: see Action Items and background here: http://wiki.geneontology.org/index.php/Ontology_meeting_2012-08-08#FOLLOW-UP:_EC_numbers_in_GO_-_feasibility_study
We need to add more info to the ontology header
Of particular importance is the various ontologies that import other ontologies. The Ontology extensions page is hard to keep up to date, people don't read it, and isn't well advertised.
I've added a simple html + png generator to owltools. This is driven entirely by the ontology header information and information computed dynamically from the ontology.
* http://purl.obolibrary.org/obo/go/extensions/x-chemical-importer * http://purl.obolibrary.org/obo/go/extensions/x-root-importer
The URL is always the ontology URI minus the ".owl"
Note these don't look like much at the moment as we don't have much metadata in the file! The output can be prettied up, and confusing OWL terminology such as "annotations" removed.
Some of the recommended properties to use:
- dc:title - short descriptions
- sc:description - longer
- dc:creator (GOC editors in most cases)
- dc:contributor (e.g. CHEBI)
- foaf:homepage - URL with more info
This should be primarily of use to ontology editors but we want to encourage advanced users to start consuming these ontologies