Ontology meeting 2013-08-01

From GO Wiki
Jump to navigation Jump to search

Attendees:

  • Paola
  • Heiko
  • Becky
  • Chris
  • David
  • Harold

(Jane and Tanya on holiday)


Minutes: Becky (see notes in green)


Do the ChEBI paper proofs need any more work?

The ChEBI paper has been published.... horray!!


Items bumped from last week

Logical definitions for GO terms referring to 'roles' in ChEBI

We need to discriminate between terms we want to define in GO vs. terms we 'care less' about - terms we can map to the ChEBI roles.

Examples of terms we lack logical defs. for so far are listed in email thread ('Logical defs. for 'role' chemical terms').

Resolution: the discrimination should be on this basis: terms we want to define in GO with respect to GO biological processes should have logical defs. referring to GO terms; other terms can have logical defs. referring to a ChEBI role term. The latter can be considered placeholders, we can change them to refer to GO if we add related process terms in GO.

AI: Paola to add xps for terms that can safely refer to ChEBI roles (e.g., 'xenobiotic catabolic process').

As for the logical definitions that we want to refer to GO, we need Chris, so discussion is bumped to this week.

Chris already was leaning towards having our own structure for some of the roles because some of the ChEBI relationships don't make sense for us (e.g. antibiotic is_a toxin). We will create our own structure ontology for this bit.

We will think about our definitions. We need to define in GO, what does toxin transport mean? Does it have to happen in the context where the transport is toxic to the organism involved. Or just where the substance has been classified as a toxin? Neurotransmitter is easier and can be defined by process context.

  • AI: CREATE A JIRA TICKET FOR MAKING OUR OWN ROLE ONTOLOGY.
    • We will Make a GOCHE-like ontology for roles, structuring how we like with our own IDs. We will need to connect to CHEBI for biochemical roles (e.g. cofactor).


Continues from last week, and beyond: Updates on ongoing work

Two directories in /scratch with initial work. Look at go-rhea-xref-cardinality.rpt, there are 75 one-to-many (about
half relate to NAD(P)H). Demo by Chris in Protege with file /scratch/rhea/mf-merged.owl.  May not need a Rhea link for
every reaction.

Chris not on the call, so bump to this week.

FIRST STEP is to make sure there are 1:1 mappings between enzyme reaction terms and RHEA terms. Becky will take a first look at these and fix where possible. We currently have 1:many, many:1 and many:many. We need 1:1.


Propagating via positive and negative regulation

Do we want to have this property chain, or could the asserted links not always be safe - or could they involve loss of information?

Discussion: it's not always satisfactory to have that chain of links asserted; we lose information about whether regulation is direct or indirect...

We may revisit the idea of 'indirectly_regulates'. We also need to play more with tests. David presented the LEGO example of glycolysis.

AI: Ask Paul Thomas to provide the full example file??

At some point we want to discriminate between direct and indirect regulation.

Dodgy inferences from ChEBI roles

see email thread 'funny inferences'

ADD GO:0042891 'antibiotic transport' GO:1901998 'toxin transport' (to the microbe, yes)

ADD GO:0071934 'thiamine transmembrane transport' GO:0006855 'drug transmembrane transport' (is a vitamin a drug?)

ADD GO:0090457 'histidine transmembrane import into vacuole' GO:0006855 'drug transmembrane transport' (even better, is an amino acid a drug)?

ADD GO:1900515 'regulation of xylose catabolic process to ethanol' GO:0043469 'regulation of D-xylose catabolic process' (change regulation term to D-xylose)?



Project management in Jira

A reminder that if we don't have time to review this today, we should all look at/update our assigned tickets.

Retrieved from "https://wiki.geneontology.org/index.php?title=Ontology_meeting_2013-08-01&oldid=52925"