Ontology meeting 2016-02-11
Attendees: Melanie, Paola, David OS, Harold, Heiko, Tanya, David H
Minutes: Paola
use of comments
Is it appropriate to use comments for link to github tickets etc? E.g., when creating a new term, it is sometimes useful to indicate why we did it this way, especially if there was long discussion on github. However comments are public, so is this ok? Do we need a field for "editors comments"?
Convenient to be able to retrieve discussions quickly, but generally agree we don’t wish such links to be visible in fields displayed by user-facing tools (e.g. comments) Could go in as a property value, set so it isn’t released to the public. Heiko: if it were in the obo file, it’d be publicly visible. DOS: we could filter that field out same as we do with some relations details. Not sure we need to though. What about using the xref field and set it to the GH tracker URL? DOS’ proposal using xref: https://github.com/geneontology/go-ontology/issues/12281 > xref: tracker: 12281 then in xref abbs yaml: tracker: https://github.com/geneontology/go-ontology/issues Heiko: should be more precise e.g.“GO_ontology_tracker” rather than simply tracker. All agree that could work.
AI: To add new stanza to xref abbs YAML. Once this is in place, all can edit. Just need to agree on name.
Melanie: a colleague wants to know whether terms like 'running', 'climbing', 'jumping' could be added to GO. Goal is to describe 'exercise biology'. We have terms such as 'jump response' and 'walking behaviour' - are we willing to expand to other similar? They are physiological processes, but at a larger scale.
Melanie’s colleague is working on an exercise ontology. It sounds like it would be focused on human activities, but we need more details. Tanya: those terms couldn’t be used for annotating gene products; Paola: I wouldn’t expect them to be, and generally speaking, we'd be ok with hosting terms for non-annotation purposes, e.g. image annotation. DOS: there is a potential for overlap with the Animal Behavior Ontology (ABO) and the NeuroBehavior Ontology (NBO). FlyBase has annotations to locomotory behaviour - researchers do screens on that. Would have to define terms in a way as generic as possible, so as to potentially be applicable to non-human species.
AI: Melanie will ask for more details on the project in general, and to submit a ticket with details of terms requests. Then, keep an eye on ABO and NBO.
Upper ontology needed?
DOS: Found an example of a GO:process regulating a pro:protein. Really should have disjoints, domain and range in place to detect things like this. In this case we need BFO:process as a parent to MF and BP; regulates domain and range -> BFO:process. Protein under BFO continuant (probably indirectly via ChEBI?) and disjointness axioms from BFO. BFO2 (classes only) should be suitable for this.
Melanie: Maybe we can have a small import file with those in for editors only, and we could at least try and do some clean-up? With the goal of having them actually enforced in the future. DOS: Good idea. Need somewhere for bridging relations up to BFO to live. Perhaps in a separate bridge file for now?
BFO has the disjoints already. We don’t have a lot of time or resources to test this at the moment, but it’s something that needs to be done… We could try it out incrementally and roll it back if it breaks stuff, until we have time to fix. Would probably use a bridge file.
AI: wait for Chris to discuss with him. If he agrees, he’ll need to do the minimal set up. Chris and Heiko will need to think about how this fits in with the pipeline.
release process
The page at http://viewvc.geneontology.org/viewvc/GO-SVN/trunk/ontology/go.owl?view=log shows release on Mon Jan 18, Wed Jan 20, Thu Jan 21, Fri Jan 22, Wed Jan 27, Fri Jan 29, Sat Jan 30, Tue Feb 2, Wed Feb 3, Sat Feb 6, Wed Feb 10 - doesn't seem to follow a pattern. Though they are nightly builds generated, see http://build.berkeleybop.org/view/GO/job/release-go/
Heiko: there is another Jenkins job called build go plus, https://build.berkeleybop.org/job/build-go-plus it relies on the makefile. Details: ## Go to the correct folder cd svn/extensions make --always-make go-plus.owl go-plus.obo # commit to svn SVN_SSH="${global_svn_ssh_commit_line}" svn commit -m"update go-plus.owl" go-plus.owl go-plus.obo # imports are made as a side-effect of go-plus SVN_SSH="${global_svn_ssh_commit_line}" svn commit -m"update imports from go-plus job" *_import.owl *_import.obo go-*bridge.obo go-*bridge.owl This job sometimes fails for some reasons. Also, there is an additional step, based at Stanford (they host svn), that’s been there for ever. That’s not maintainable in the long term, so there is an open GH ticket assigned to Heiko about streamlining the process. DOS: are there more steps where we could have warnings in place? Heiko: not easily feasible at the moment due to the structural issue above.
Fixing relations import
The OWL file needs two labels for round-tripping via OWLtools to work!
Melanie: Relations need to have a label in the GO file (in addition to the one they already have from RO); we sent the stacktrace by email.
AI: create a ticket; Heiko will look into it.
Some relations were lost and then re-appeared in editors file - need to work out why
Following investigation and discussion we think this happened when Melanie updated the trunk while editing OWL file, then tried to use Merge to recover. Stemming from discussion: DOS put together some useful documentation about how to safely recover from this using mac merge tools: "Here's a quick doc on easy 3-way merge and why you might need it. https://docs.google.com/document/d/1fOTm_CdrrrblqwEHWaZ6lMZFDa5LCTbmkA8MF8PfAkc/edit?usp=sharing Comments please. Once we're happy with it I can use PanDoc to convert to wiki and post on the GOC wiki."
AI for all: send feedback to DOS. AI for DOS: post on wiki and copy wiki link here.
Follow-up from Harold
Harold: met with PRO; resolved to actively import Panther families in PRO. All agree that’ll be useful.
AI: Harold to keep Paul T posted on this.