Ontology meeting 2013-06-20

From GO Wiki
Jump to navigation Jump to search

Attendees: Harold, Chris, David, Heiko, Jane, Judy, Paola, Tanya, Becky

Minutes: Paola + Becky

CHEBI paper

AI: We all need to look at CHEBI paper in the Dropbox folder (GOCHE_fina-postreview). Let David know if we have any concerns. He will send to all other co-authors tomorrow.


New error report (from Heiko)

When the GO build fails, OORT now creates a module file related to the error, to allow easy debugging in Protege. This will be a small file in owl that we can quickly load in Protege to see where the problem is. The permanent link to where the file lives in the dashboard is:

http://build.berkeleybop.org/job/build-go/lastFailedBuild/artifact/validate/staging

The file is called go-unsatisfiable-module.owl

Load it in Protege, then use Hermit to see the explanation.


Cycle error over has_part/part_of

Chris: the links that Jane suggests are valid, but we haven't encountered this situation before... Tools that consume the full GO (such as QuickGO) may have problems with it.

David: create new relationship 'necessarily_part_of' that includes both part_of and has_part?

AI: We need to check with Tony. QuickGO should stop assuming that GO is acyclic. Jane will put in a ticket.


FOLLOW-UP: New TG templates

Paola has tested the cell_apoptosis template and it's ready to advertise.

Are the plant ontology templates ready to advertise too?

AI: Tanya will advertise. (DONE)

New terms for Giardia annotations?

Main issue: those terms are defined spatially - is this more in scope for anatomy or GO CC?

The alternative would be to help them to start a new ontology in owl, that imports GO too.

After brief discussion, we agree to (look into) putting them in.

AI: David will look into this (how general they are) after he's done with apoptosis re-curation; he'll email Rama about it. Chris might then look into automating at least part of the process.


BUMPED FROM LAST MEETING: GO is_a based on ChEBI has_role? (Tanya)

Bringing back the discussion from past meeting. Recent SF request from Peter dE brings it back up . SF item

We would need to make the specific 'chemical as vitamin' term. For example, nicotinamide metabolism does not always occur in the context of a vitamin.

To keep it safe, we should create specific children (e.g. nicotinamide vitamin metabolism, as an instance of nicotinamide that works as a vitamin, whatever that means; specify that in the definition itself; is_a water soluble vitamin...). However...

One suggestion from David is to bump this to column 16. This wouldn't work for mappings (reactome) though.

Chris voiced the concern that generating child terms for all cases (nicotinamide vitamin metabolism etc) is ontology purism gone too far. So we looked at the annotations that were made to nicotinamide metabolism and biotin metabolism terms. For both, PomBase, Wormbase etc had annotated to these terms. And Val hasn't yet complained that they're breaking the true-path rule.

So, for now we're going to make nicotinamide metabolism is_a: vitamin metabolism, as Peter requested. (DONE)

BUMPED FROM LAST MEETING: Missing EC numbers in GO => next week

It seems we don't have the resources to implement a pipeline for these. Shall we wait and see if the new GO Editor can take this on? Shall we discuss this at the next GOC meeting to see how much of a priority it should be?

In the meantime, we can add single terms when requested by curators, cross-referencing to Rhea or asking Rhea to add the reaction if missing.

But can we close the following SF tickets?

https://sourceforge.net/p/geneontology/ontology-requests/9319/ (Becky's - nearly done anyway)

https://sourceforge.net/p/geneontology/ontology-requests/9320/ (Jane's)

https://sourceforge.net/p/geneontology/ontology-requests/9321/ (Paola's)


We never really got to the bottom of handling multi-step reactions. If we create separate GO terms for the subreactions with HAS_PART relationships to the parents, it's tricky to name the sub-reactions. There's no place at the moment where we can take these names. So, do we want to continue just to add one GO term for the overall reaction, stating in the definition that the reaction is multi-step?


BUMPED FROM LAST MEETING: Follow-up: template for assembly/disassembly of CC terms => next week

Chris/Heiko, where are we with this?


BUMPED FROM LAST MEETING: Organelle vs. macromolecular complex => next week

Stemming from some observations on inferred links in the assembly node, e.g.:

(Added inferred link) GO:0000921 'septin ring assembly' GO:0070925 'organelle assembly'

GO:0000921 'septin ring assembly' also has the is_a parent 'cellular protein complex assembly'. This was added back in 2008 (http://www.ebi.ac.uk/QuickGO/GTerm?id=GO:0000921#term=history). However, 'septin ring' only has 'organelle' as ancestor, not 'protein complex'. Should we make 'macromolecular complex' and 'organelle' disjoint?

The proposal would be to have

organelle DisjointWith mm complex


BUMPED FROM LAST MEETING: 'Encoded' vs. 'non-encoded' molecules in GO => next week

Stemming from this SF request from Peter: https://sourceforge.net/p/geneontology/ontology-requests/10200/

At the time of writing, I'm waiting for Peter to suggest defs. and placement.


BUMPED FROM LAST MEETING: Muscle Fiber vs Muscle cell => next week

[1]

Also see email thread from Chris (subject:muscle fibers in a tangle)

Muscle Issues

  • We need to go through each fiber term and check whether we mean the contractile fiber (CC) or the muscle cell (CL).
  • Should we use the CL naming as default?
  • GO isn't consistent in CC and BP about naming.


BUMPED FROM LAST MEETING: Project Management in JIRA => next week

As we likely won't have time to go through this, and we won't have an Editors call next week due to EBI GO Editors interviews, may I suggest that we all look at our tickets and schedule progress on them if necessary.


MEMBRANE SIDES VS LEAFLETS => next week

SF item from Alan Bridge: https://sourceforge.net/p/geneontology/ontology-requests/10223/

Is a membrane side equivalent to a membrane leaflet (one of the lipid layers)? If so, can all 'internal/external side of membrane' terms be changed to 'internal/external leaflet of membrane'?

See:

  • leaflet of membrane bilayer ; GO:0097478
  • external side of plasma membrane ; GO:0009897
  • internal side of plasma membrane ; GO:0009898
  • GO:0031232 extrinsic to external side of plasma membrane
  • GO:0031233 intrinsic to external side of plasma membrane
  • GO:0031234 extrinsic to internal side of plasma membrane
  • GO:0031235 intrinsic to internal side of plasma membrane
  • GO:0031237 intrinsic to external side of plasma membrane, in periplasmic space
  • GO:0031362 anchored to external side of plasma membrane
  • GO:0071575 integral to external side of plasma membrane

TG template for 'cell differentiation'

Are we ready for this?

See e.g. http://sourceforge.net/p/geneontology/ontology-requests/9698/


TG template for 'anatomical structure development' => next week

See e.g. http://sourceforge.net/p/geneontology/ontology-requests/9697/

We're probably not ready for this yet, but we may want to schedule necessary work - or push to a later date?


STRUCTURAL MOLECULE ACTIVITY => next week

Is there a long-term plan for where we can't use annotation extension? E.g. for mappings. Looking for a term to map to KW: viral matrix protein, which is a structural constituent of virion.