Ontology meeting 2014-05-15

From GO Wiki
Jump to: navigation, search

Attendees:Tanya, Harold, DavidOS, Chris, Heiko, DavidH, Becky, Jane, Judy


Adding GOTAX IDs to taxon relationships created in editors file

Right now, taxon rules added in the editors file are not assigned a GOTAX ID, and hence NOT picked up by checks or visible in GO browsers (QuickGO currently). We've been adding a number of taxon rules lately and it would be good if they could actually be applied :-) There is a Jira ticket for this but it's unassigned, could the priority be upped maybe? Thanks!


QuickGO requires there to be an ID for every rule, so when we add a relation in OE we need to run a script to create the rule. Can we have the script run on Jenkins nightly? The current script will need some modification. Chris will modify the script.

Follow-up: Editing in Protege

Tanya posted the guidelines here: http://wiki.geneontology.org/index.php/Editor_Guide#Editing_in_Protege. We should have all tried them out by now. Report.

(To start with, we should at least try the roundtrip and make sure we get nothing in the diff.)

  • Harold can convert the files, but cannot load them into protege. Presumably a problem with catalog.xml file
  • Jane still to test
  • DavidOS works fine (there is a save issue if the ontology is open for a long time. Be careful not to edit anonymous ancestors)
  • DavidH works fine
  • Becky not yet
  • Paola not on call
  • Tanya works fine

Cell motility design pattern

See email thread with same subject. Stemming from https://sourceforge.net/p/geneontology/ontology-requests/10719/

This is a miniproject in itself. We need to be able to classify the types of cell motility.

Inferring parts

What needs to happen for us to be able to do this?

Can we add a GCI at termgenie addition? Would they be added one time en mass? There are some TG templates that already deal with part_of relations. Maybe we just need to go with a simple rule approach.

Integrating UBERON

Chris and DavidH worked on vetting the anatomy logical defs last week in CA. Are we ready to go with those? See go_uber.obo in the editors directory.

Warning for terms obsoleted/merged in imports

I just noticed that chebi has merged 'secondary metabolite role' (http://purl.obolibrary.org/obo/CHEBI_26619) into metabolite. We use this in GOCHE, but the loss of the term in the import file has not, AFAIK, raised any flag for use to fix our terms. Is this a general issue?

It would be good to have a process in place to know when things happen at ChEBI. So what do we do about secondary metabolism? Get Paola or Jane to follow up about secondary metabolite with CHEBI.

Inference from CHEBI roles

It is probably a good idea not to make full use of inference from biological role as asserted by CHEBI - but we may still want to in some cases. Right now, these inferences are blocked, but for rather obscure reasons:

1. We make intermediate terms like 'substance with metabolite biological role' EquivalentTo 'chemical entity' and (has_biological_role some metabolite).

2. These intermediate terms are used in the definition of GO terms such as: secondary metabolite EquivalentTo 'metabolic process' and ('has participant' some 'substance with secondary metabolite biological role').

3. We get no inference from CHEBI has_role assertions because all of our GOCHE terms are defined using SubProperties 'has role', such as has_biological_role. These have GOCHE identifiers so do not live in RO and are not available for CHEBI to re-use.

In discussion earlier this year, Chris suggested that, where we do want inference, we should add GCIs of the form:

chebi:has_role some 'secondary metabolite' SubClassOf go:has_biological_role some 'substance with secondary metabolite role'

Should we start adding GCIs? Or can we find a more obvious way to fix this?

Chris also suggested: "The chemical roles we may want to take across en-masse; these are more like qualities." Should we attempt this, if so how?