Ontology meeting 2013-09-05
Attendees: Jane, Paola, David OS, Heiko, Becky, Tanya, Chris, David H, Harold
Minutes: Tanya
happens_during (carried on from last week)
What's the next step for this? There's been some confusion on the email list.
We talked about using this relation between phases and regular processes, but it seems it's already being used in c16 between pairs of regular processes. Are phases types of processes? Can we finalise the definition of happens_during?
- David OS has file which collects notes about relations with info about Allen relations David OS' doc
- happens during = seems to be union of Allen relations during, finishes, finished by
- might be a naming issue, Allen relations are not intuitive
- Jane will bring up Allen relations in her discussion with Val next week about happens_during and cell cycle
Equivalent classes for Acid/Base transport and transport activity (bumped from last week)
During the merge of the x-transport the following inferred equivalent classes came up:
GO:0008519 'ammonium transmembrane transporter activity' GO:0051739 'ammonia transmembrane transporter activity'
"In this review we use the term ‘ammonia’ to refer to the combination of NH3 and NH4+. When referring to a specific molecular form, we will specifically identify either ‘NH3’ or ‘NH4+’.(snip) Ammonia exists in two molecular forms, NH3 and NH4+. Renal ammonia transport has been thought classically to involve passive NH3 diffusion and NH4+ ‘trapping’. However, both NH3 and NH4+ have limited permeability across plasma membranes and are transported by specific mechanisms."
GO:0015708 'silicate transport' GO:0051207 'silicic acid transport'
"The term ‘‘silicon’’ not only refers to the element, but is used as a generic term when the specific form of a silicon compound is unknown [17]. The predominant form of silicon in aqueous solution at low concentrations is silicic acid, Si(OH)4. This is a weak acid with a pKa of 9.8 for the formation of SiO(OH)3-."
Both pairs are correctly defined with the appropriate ChEBI classes. Due to our bio-chebi adapter, we infer them to equivalent. Tanya mentioned that some transporters might be specific to certain forms so it might not be possible to merge them
- and we have to rethink our entire strategy for conjugate bases....
Harold will investigate and report back. All: think about possible solutions.
Defining complexes by their members
See email thread from Chris.
So we have >200 complexes defined by what they do (thanks Harold!)
What is our plan for the remaining ~1300 complexes most of which would be defined either by members or some combination of members, location and function?
Where do member-defined complexes live? GO? PRO? Can annotation to member-based complexes be automated? What good does it do GO to have member-defined complexes? Useful if we can link that term to its function/process. What is most important to address? Links to process/function more impt than link to component. Do some data analysis- which ones are species-specific.
Flybase nomenclature link http://flybase.org/static_pages/docs/nomenclature/nomenclature3.html#11.1.
EC2GO mapping
File has not been produced in ages. Heiko will find stop gap measure until Rhea solution is found.
NEW TG TEMPLATES
Where are we up to for the latest templates:
- protein localization
- cell differentiation (https://www.ebi.ac.uk/panda/jira/browse/GO-181)
- ask Chris about missing xps in the x-cell.obo file OBO file is the last one that has actual xps. Where did they go?
- Chris: They are in the live file but the xps are incomplete, not all cell types exist in an external ontology. Possibly 150 new CL terms to add. Add to CL or add to Drosophila anatomy ontology?
- CC assembly/disassembly (https://www.ebi.ac.uk/panda/jira/browse/GO-177)
NAD(P)
How much do we want a 1:1 mapping between GO terms and RHEA terms?
With exception of NAD(P) we will strive for 1:1 mappings. For the NAD(P) ones we make the GO reaction equivalent to the union axiom. There's also a (d)CTP example which might fall in the same category but not many.
We've been in touch with Alan Bridge about the NAD(P) reactions, and whether RHEA could create a union reaction for these GO terms (which currently have 2 RHEA mappings)
He says......
I consulted with Kristian Axelsen (CCed here) on the usage of NAD/NADP in IUBMB reaction definitions (he is the only IUBMB NC member I could get hold of).
We think the text the IUBMB provide may need rectifying slightly to something like this:
"The use of NAD(P) indicates that NAD or NADP may be used by the SAME enzyme with similar affinity. In practice, that means less than 10 fold difference in activity between the two."
So the intended meaning seems to be "NAD(P) means NAD OR NADP", but not that the reaction contains NAD OR NADP (and we don't know which), but rather that the enzyme catalyzing the reaction accepts NAD OR NADP - the enzyme will use either - which means AND! (in terms of protein function annotation)
It's hurting our heads. How does this affect our GO terms and mappings?
Assert-Report
Anything to discuss this week?
- Fairly straightforward. Last two weeks also looked ok with nothing strange.
- Some marked inferred, not entailed links are not completely gone yet, some keep coming back.
Col 16 workshop in BH
Good for ontology developers to attend. Clarify happens_during vs. overlaps/during (Allen relations). Some Jenkins checks could be useful.