Ontology meeting 2012-11-29

From GO Wiki
Jump to navigation Jump to search


ATTENDEES: Paola, David, Tanya, Harold, Chris, Heiko

Children of 'translation' causing TPVs

Stemming from this SF item: http://sourceforge.net/tracker/?func=detail&aid=3590949&group_id=36855&atid=440764

Fix: remove CHEBI xps for the offending terms only.

Also look at any xp that has CHEBI 'protein' in it.

AI: David/Tanya: work on the xps to fix Val's problem, and comment on the SF item.

AI: Paola and Jane to discuss the terms highlighted by Val with CHEBI next week.

Chris says we can prevent these TPVs from occurring using disjointedness axioms.

But the issue should be addressed anyway if CHEBI revises its protein terms/links.

Similarly for toxins (see Val's other SF item: http://sourceforge.net/tracker/?func=detail&aid=3590771&group_id=36855&atid=440764).

Do we then need NOT to reason over CHEBI roles? Because this causes many wrong inferences...

One solution may be to have our own role ontology, similar to goche, use-case driven (e.g. response to DDT is_a response to insecticide where the is_a link is hardcoded)

AI: David: Remove the is_a link between response to organophosphorus and response to insecticide

ChEBI paper?

All done?

AI: Chris to follow up on Jane's email about flow diagrams.

AI: David to prepare cover letter.

AI: Jane and Paola: check with CHEBI that it's ok to proceed to submit even without approval from Paula Matos and Steve Turner. DONE, it's ok to go.

Documenting decisions

We need to develop a SOP for doing this, otherwise we won't do it. Let's take the example of the metal cations from last week and figure out how we should do it. We should get into the habit of when we have come to a decision like this, making a ticket for someone to make the doc.

Wrt metal ion terms - How to handle metal ions from the CHEBI changes last week? Change term names to be 'cation' or keep as ion and implement as a GCI? Jane has already changed the xps to the CHEBI cation entities.

To be discussed at the next call (or earlier by email if needed).

Other examples?

Adding has_parts for antiporters

Arose from this SF ticket: [1]

Essentially Ruth wants to infer both calcium export and sodium import from calcium:sodium antiporter activity. I've suggested we could use has_part, but this is probably something we'd want to do globally. Do we want to do this? What would be the implications? Could we extend to other transporter types? We have transporter part_of transport elsewhere...this is kind of the inverse of what's being suggested.

FOLLOW-UP: Endosomal transport

All done from last week? (http://wiki.geneontology.org/index.php/Ontology_meeting_2012-11-08#Endosomal_transport)

Plan was: create vesicle-mediated endosomal transport as of endosomal transport, put all current kids under this (parent can be removed from vesicle-mediated parentage), send email to dbs to either move or leave in place, still correct

Did this: removed vesicle-mediated transport (VMT) as parent of endosomal transport, revisited all kids, copied those that referred to transport using vesicles to VMT. No new term created. Annotators can annotate to 'endosomal transport' or more specific kids, as they see fit.

Chris created xp for endosomal transport with new relationship:

  id: GO:0016197  ! endosomal transport
  intersection_of: GO:0006810 ! transport
  intersection_of: results_in_transport_to_from_or_mediated_by GO:0005768 ! endosome
  id: results_in_transport_to_from_or_mediated_by
  name: results_in_transport_to_from_or_mediated_by
  comment: This relation is so far only used for one class, 'endosomal transport' (GO:0016197), which groups 
  together disparate roles the participants can play - it is defined as 'The directed movement of substances into, 
  out of *or mediated by* an endosome...'
  is_a: has_participant

involved_in TG template

Can we extend this to include MFs involved_in BPs?

TG restrictions (in view of 'global' TG tool)

David and Paola suggest to prevent submitters from creating regulation terms if the target term is a descendant of 'regulation of biological quality'.

E.g. 'regulation of maintenance of presynaptic active zone structure' is wrong because 'maintenance of presynaptic active zone structure' is a grandchild of 'regulation of biological quality'.

TermGenie free-form testing to be reported on at the next call.

Daily digest emails for new/edited/obsoleted terms

Could we have those back please?

AI: Heiko and Chris to look into it.

Alerting users of long-term changes to the ontologies

Peter D'Eustachio emailed recently asking about some comments Amelia added to some catalysis terms alerting that they would be made obsolete. It lead to a discussion of whether we could provide a mechanism for our users to be alerted about upcoming changes to the ontologies, longer scale than the obsoletion notices.

I know it's something we've discussed at GOC meetings in the past, but we've never really settled on anything. The UniProt group do something ad-hoc in QuickGO where they mark some terms as being under discussion e.g. [2], but it's pretty hacky and uses reg exp.

FOLLOW-UP: Capturing explicit cell lineage in GO

See background here: http://wiki.geneontology.org/index.php/Ontology_meeting_2012-11-15#Capturing_explicit_cell_lineage_in_GO

To be discussed later on.

TermGenie paper?

Plans? Submit abstract for Biocuration 2013?

We should start work on this, and maybe present an abstract at the Biocuration 2013 conference.

Follow-up on annotating to 'part' terms (nuclear_part, etc)

Jane has already added the part terms to the 'do not annotate' anti-slim. Chris is adding a check to Jenkins, so that's everything for now; we'll just wait to see what the check throws out.

Come back to this at the next call if needed.