Ontology meeting 2015-03-05

From GO Wiki
Jump to navigation Jump to search

Attendees:

Minutes:


Acids and bases in GO

Stemming from https://sourceforge.net/p/geneontology/ontology-requests/11524/

The solution we've adopted so far is proving unsatisfactory specifically with respect to NH4+, NH3 and NH2- conjugate-acid-base relations. We need to think of an alternative, with or without the direct involvement of ChEBI.

Does NH2- exist in biology (azanide ion) (Harold: I can't find; it is really unstable; reacts with
water quickly to form NH3 + OH-; Azanides as a class: Chebi has; these aren't like natural metabilites
More like drugs, etc. 
Option: exception to the GCI generation rule (equivalence is transitive so we have to do this): 
for azanide; We should document this somewhere very well for all times we have to do this.

Update from Heiko (March 10th): "as discussed on the call, we have modified the generator for bio-chebi. It now supports exceptions for the generation of the GCI axioms, i.e. ammonia and it's conj base/acids. The current approach to define these exception is a new subset in the seed file for bio-chebi generation called no_conj_equiv. The bio-chebi.owl update file has been committed today. "

Rhea plans

Background and initial plan: https://www.ebi.ac.uk/panda/jira/browse/GO-190

Paul will report with his discussion with Rhea
Swiss Lipids project: Potential source of annotations, have large number of enzymes operating on
lipids with evidence, etc.  could convert to GO if had a mapping of MF to Rhea. 
We need an expert in the field to straighten out our lipid-related terms. 
DPH cross references to non-directional reactions in Rhea (no order in a GO pathway)
TB: Metacyc mapping question?: DPH cross references Metacyc; try to cross reference to 
Kegg, Metacyc, Rhea; BUT: don’t want automate.
Chris had tried looking at Metacyc to Chebi.; what about EC mappings?
http://sourceforge.net/p/geneontology/ontology-requests/10687/
Make a list of rhea terms that DON”T map to EC


Do we have resources for this? Chris, David H, Harold?

How to define protein families in enzyme binding terms

This follows up from a previous call where we resolved: "We have decided that we will include families. How we define families will need to be addresses. Is it only if there is more than one similar gene in the same organism? The bottom line here is that a gene protein is not binding the activity, it is binding the molecule."

We need to address how to define families.

Just use protein binding. We can use the “inferred from” field to group if needed.
Move them to IDA with annotation extension to indicate the partner

Replacing has_regulation_target with more specific, chain based relations

In discussion with annotators (Ruth, Rachael, Becky, Aleks) - agreed that the confusing annotation extension relations - has_regulation_target - should be replaced by more specific relations defined using property chains. We already have some of these, but the names, which embed the property chain, are confusing to most people.

What relations to have and what to call them?

regulates_o_has_agent -> regulates_activity_of - with restriction that only applies to regulation of MFs (BPs having many agents) ? regulates_o_has_substrate -> ?

How deep to go?

DOS: What do we do with existing has_regulation_targets: merge to regulates_has_participant”; 

regulates_transport_of ? regulates_transcription_of ? <- This might seem a step too far, but it would be much clearer than the current confusing mess arising from using has_regulation_target

To what extend should we define these relations for GO vs generically?

e.g.: regulates_o_has_agent:

  • generic: A relationship between the regulation of a process and the agent that executes that process.
  • GO specific: A relationship between a molecular function or biological process that regulates a molecular function and a gene product that executes the regulated function.

Should we add both direct and indirect regulation varieties? Note - potential confusion here between directness re process/fn vs directness of regulation.

DH: This isn't really that confusing. All of the regulation terms in GO except for the regulation of biological quality refer to GO processes or molecular functions. The regulation chain relations are simple chains of the regulates relation with whatever relation would be used either asserted in the ontology or in the logical definition of the term. As a simple example 'regulation of protein kinase activity' would be the chain regulates_o_something. The something would be any value that can be used in the description of the kinase activity. It could refer to the kinase itself, regulates_o_has_agent (if this is the relationship between the molecular function and the annotated gene product) or it could be regulates_o_has_input to describe the protein that is being phosphorylated, or it could be regulates_o_has_output to describe the specific phosphorylated protein. The key here is getting the relations of the root process and function terms correct to describe things that are going into processes and coming out of processes. That all ties together getting good axioms in the ontology with the annotation extension relations. At the end of the day, to get proper folding and unfolding, all of these need to align. So there are two challenges; defining and determining the correct relations between entities and the primary processes and coming up with names that curators can easily use. The current chains are not user-friendly.

DPH: sometimes we use gene id sometimes UniProtKB id; that’s fine

results_in strategy (bump discussion to next week)

See emails from Chris following last week's call (threads on go-ontology 'results_in outliers' and 'results_jn refactor'). Chris asked for feedback on his proposal. Discuss, then reply to email thread.

results_jn refactor: Looks OK, but has many very indirect cases.

BTW, Antonia recently requested this on TG: 'positive regulation of pyrimidine-containing compound salvage by positive regulation of transcription from RNA polymerase II promoter'. Put on hold until we decide... (UPDATE: Paola obsoleted the TF freeform request and committed it using the (old) reg_by_reg template, for the time being, so Antonia can annotate.)

Transporter activity part_of Transport process

Paola reminded us of GO_REF:00090. This process is currently manually triggered (also not a Jenkins job at the moment). Currently, there were five missing part_of links:

 GO:0061459 'L-arginine transmembrane transporter activity' 'part_of' GO:1903400 'L-arginine transmembrane transport'
 GO:0015093 'ferrous iron transmembrane transporter activity' 'part_of' GO:1903874 'ferrous iron transmembrane transport'
 GO:1903425 'fluoride transmembrane transporter activity' 'part_of' GO:1903424 'fluoride transmembrane transport'
 GO:0005304 'L-valine transmembrane transporter activity' 'part_of' GO:1903785 'L-valine transmembrane transport'
 GO:0015208 'guanine transmembrane transporter activity' 'part_of' GO:1903716 'guanine transmembrane transport'

David added the links.

Left to work on:

1) If a more stable solution is not near in time, would it be possible to schedule Heiko's script to run say once a month or every six months in the interim?

2) (Minor) Edit GO.references to correct this in GO_REF:0000090:

"'transporter activity' and 'transports_or_maintains_localization_of' some X' -part_of-> "transporter and 'transports_or_maintains_localization_of' some X"

should read

"'transporter activity' and 'transports_or_maintains_localization_of' some X' -part_of-> "transport and 'transports_or_maintains_localization_of' some X"

[Paola] Can I edit this via SVN, or does the 'real' file live somewhere else. [Now done, via SVN]

Function-process links

Rama reminded us that we should reply to this user [now done]:

https://mailman.stanford.edu/pipermail/go-discuss/2015-February/007159.html