Ontology meeting 2016-03-10

From GO Wiki
Jump to navigation Jump to search

Attendees: Paola, Chris, Tanya, Harold, Judy, Paul, DavidH, David OS, Melanie

Minutes: Melanie


GO slims

[Melanie] I'm looking at updating the metagenomics slim in collaboration with InterPro, and was wondering how slims were being generated?

The page at http://geneontology.org/page/go-slim-and-subset-guide mentions "The GO slims in this list are maintained as part of the GO flat file. The files available below for download are generated by script from that file." Does anybody know where is that script? I'd rather not rewrite something that's already out there

Also the page states that those slims are maintained, but looking at the metagenomics one and for example the PIR one, they both contain obsolete terms. Is this something we should be keeping an eye on, and if yes should we perhaps have a check in the slim generation script to prevent this?

Finally, how to fix? I don't know what is the policy for updating those. Easiest would be to remove the obsolete terms, but do we need to check with the original creators of the slims?

[There's been some discussion on the mailing list, but looks like a few questions may still be pending. Melanie/Chris?]

Minutes

  • Some slims may contain obsolete terms
  • For archived slims: nothing to be done
  • For current slims: we can keep them there, but annotations should be filtered out. As label was modified to include ‘obsolete’ shouldn’t be an issue.
  • who is responsible for the slims? Judy: we should add MGI slims. Perhaps Mary would look at it

* Chris: we should add a created date, last modified date.

  • go_slim is actively maintained, add purposes, add list of others slims below saying ‘some people have found those useful at some point, but they may not be complete or current even though they’re generated on a daily basis’


Cell Cycle Meiotic versus Mitotic

Currently there are a number of terms under the mitotic cell cycle that are both named and defined as cell cycle terms without specifying whether they are mitotic or meiotic. I think we need to work with Val to sort all of these out. There are two option, to make the ones that don't have meiosis- and mitosis-specific children all mitotic, or to move them up and make them generic, which will make annotations less specific. I am hoping Val will know if the ones currently in place all have counterparts in both cell cycle types.

Minutes

  • a meiotic term was under the mitotic. David OS suggests to add disjoints, but many more violations in current file. History: when we add the cell cycle meeting, it was decided to only develop the mitotic side and would look at meiosis later. Specific mitotic terms were created knowing that the meiotic ones would need to be filled in later. Under mitotic cell cycle, all the generic cell cycle checkpoints have been added as children. DavidH doesn’t know if those checkpoints are mitotic specific or not.

* AI: David H will check and update/create as needed.


ETINES

Progress:

  • Number of ETINES reduced to 1405
  • 377 flagged as delete
  • 43 flagged as keep - manual
  • 65 flagged as keep - automate. 25 of these fixed so far (will disappear on re-reasoning)
  • A further 376+ flagged with a ?. These are mostly cases where we need a general policy. e.g. there are 316 ETINES under single organism X terms.
  • 419 yet to be properly reviewed (some of these have comments)

Review plans:

Have moved to Google spreadsheet (https://docs.google.com/spreadsheets/d/1bSkgMxMHm5Aqw6UuTzTNXuTbjVZ7Uvuyos4VrLk5y4c/edit#gid=0). I will periodically freeze to update.

Sharing out work: Everyone try to do 150? Note - I've also flagged people in notes.

Minutes

  • ETINES [high priority]
  • things that were entailed and not longer are (EXISTS, TAGGED-INFERRED, NOT-ENTAILED). This is a one off - once done we switch to purely automated system.
  • David OS going through GO projects ‘archeology'
  • in ETINES spreadsheet:
  • relationships that are no longer entailed. Once upon a time reasoner inferred that ‘corticosterone binding' was a type of type of alcohol binding. ChEBI doesn’t classify steroid as alcohol anymore so we don’t get that anymore
  • status = 0 means delete it (e.g., we trust ChEBI to be right when they change the assertion)
  • strategy: try and understand why the inference is not entailed
  • goal: move to ‘flush and replace’ system so that we don’t have to manually update anymore

* AI: all to check and try to clean up a bit


Neurotransmitter role

Are a ligand/receptor pair and NT/NTR pair even when they act outside of the nervous system

https://github.com/geneontology/synapse/issues/65

Minutes

  • even if the protein is still called “neurotransmitter receptor’, their function obviously only occur in the nervous system. How do we get that context information?
  • use part_of some ‘neurotransmission’: maybe too strict? People use the term more loosely, sometimes thing being part of neuromodulation are still called neurotransmitters. AI: use part_of ‘synaptic signalling’
  • use occurs_in synapse/neuron/post-synaptic membranes: a bit more flexible. David H: are immunological synapses under synapse? DavidOS: no.
  • use chebi neurotransmitter role.

* AI David: check the way dopamine/dopamingeric receptor was done.


cytoplasmic transport

https://github.com/geneontology/go-ontology/issues/12323

-> cytosolic transport? Use of occurs in?

Vesicular transport

vesicular transport proposal

Proposal for new relation - allowing merge of 'vesicle transport' and 'vesicle mediated transport' branches:

https://github.com/geneontology/go-ontology/issues/12317#issuecomment-194775043