transporters

[dosumis]

From @dosumis on March 17, 2016 11:20

I'm sure this issue has been discussed many times before. Coming up again because it is causing confusion in the synapse project.

In general usage, 'transporter activity' involves active transport:

e.g. see https://en.wikipedia.org/wiki/Ion_transporter https://en.wikipedia.org/wiki/Neurotransmitter_transporter https://en.wikipedia.org/wiki/Dopamine_transporter

For passive transport (down a concentration gradient), the term 'channel' is used.

These are the uses that electrophysiologists/neurobiologists expect.

But in GO we have

transporter activity (1089) . transmembrane transporter activity (938 subclasses) . . active transmembrane transporter activity (427 subclasses) . . . active ion transmembrane transporter activity (0 subclasses!) . . passive transmembrane transporter activity (164 subclasses) . . channel activity (162 subclasses)

Issues:

• Given normal usage, I have a strong suspicion that much annotation to the 347 'transmembrane transporter activity' subclasses not under the active or passive branches, are actually active transport. Is the active/passive distinction pushed down a level because annotators want and MF term even if they don't know whether transport is active or passive? Wouldn't the process class be enough in that case? Given this should we consider moving this differentium up?
• Many classes that should be under 'active transmembrane transporter activity' or 'active ion transmembrane transporter activity' are not.

Taking calcium ion transporter as an example:

'calcium-transporting ATPase activity' is not classified under 'active transmembrane transporter':

Immediate fixes:

• [x] 'hydrolase activity, acting on acid anhydrides, catalyzing transmembrane movement of substances' SubClassOf 'active transmembrane transporter activity'
• [x] 'ATPase coupled ion transmembrane transporter activity' SubClassOf 'active ion transmembrane transporter activity'

' calcium:cation antiporter activity' is not classified under 'active ion transmembrane transporter activity'

https://en.wikipedia.org/wiki/Antiporter

antiporter activity: "Enables the active transport of a solute across a membrane by a mechanism whereby two or more species are transported in opposite directions in a tightly coupled process not directly linked to a form of energy other than chemiosmotic energy. The reaction is: solute A(out) + solute B(in) = solute A(in) + solute B(out)."

Ditto symporter.

Fix:

• [x] add 'ion antiporter activity' SubClassOf 'active ion transmembrane transport activity'
• [x] add 'anion:cation symporter activity' SubClassOf 'active ion transmembrane transport activity'

But - in this case we may need to watch out for incorrect inferences that may result, given that only one of the transported ions is actively transported in these cases.

Relevant historical discussion: http://gocwiki.geneontology.org/index.php/Tranporter_Activity_Meeting_Notes

Copied from original issue: geneontology/go-ontology#12347

[dosumis]

Immediate issues fixed manually (see checklists above)

• Worth discussing the terminology issue around transporter? @ukemi @hdrabkin - any comments?
• Better axiomatisation would help a lot in ensuring more complete classification in this branch.
  • For this to work for antiporters/symporters we may need active and passive transport subproperties of 'transports or maintains localization of'. @cmungall - any comment?
  • Note that we are following a simple multi-inheritance pattern for compound MFs, rather than using has_part. So 'ATPase coupled transporter activity' is_a 'ATPase activity' AND is_a 'active transporter activity'. This clashes with the suggested refactoring of compound MFs using 'has part'. We should make a decision on which route to take for patterns in future. I'm now tending towards this simple multi-inheritance pattern. The main disadvantage is that it severely limits the addition of disjointness axioms. @cmungall - any comment?

[dosumis]

antiporter activity: "Enables the active transport of a solute across a membrane by a mechanism whereby two or more species are transported in opposite directions in a tightly coupled process not directly linked to a form of energy other than chemiosmotic energy. The reaction is: solute A(out) + solute B(in) = solute A(in) + solute B(out)."

But subclasses typically follow a pattern like this:

calcium:cation antiporter activity: "Catalysis of the transfer of a solute or solutes from one side of a membrane to the other according to the reaction: Ca2+(in) + cation(out) = Ca2+(out) + cation(in)."^

From wikipedia: In secondary active transport, one species of solute moves along its electrochemical gradient, allowing a different species to move against its own electrochemical gradient.

Problem: Is it possible to distinguish which of the two solutes is being actively transported in these cases? This seems necessary if we are to have terms like 'calcium-ion active transmembrane transporter activity' (requested by synapse project annotators). CC @ukemi @ValWood @hdrabkin

[cmungall]

For this to work for antiporters/symporters we may need active and passive transport subproperties of 'transports or maintains localization of'. @cmungall - any comment?

This is probably most straightforward.

Implicit in this is that there is a certain subset of MFs that will be defined using a reaction pattern (or at least have subClass of axioms, as previously discussed) and a subset will be defined using similar patterns as for BP (transporters, kinases, symporters).

[dosumis]

MF refactoring discussion:

Treat antiporter as a compound function - this allows us to deal with co-reference issues.

voltage sensing: has_part some voltage sensor activity (TBA)

[pgaudet]

This issue was moved to geneontology/go-ontology#16978