geneontology / go-ontology

Source ontology files for the Gene Ontology
http://geneontology.org/page/download-ontology
Creative Commons Attribution 4.0 International
216 stars 39 forks source link

Split and change labels ABC transporters and P-types transporters (was ABC transporters and P-types transporters? ) #17373

Closed pgaudet closed 3 years ago

pgaudet commented 5 years ago

Hello,

For the MF refactoring of transporters: We have two groups of terms that represent the same function, ATPase-coupled transporter activity, but with slightly different mechanisms: describe ABC transporters and others P-types transporters ('phosphorylative mechanism').

Is this distinction necessary ? This seems to group protein families rather than different functions.

@thomaspd @amandamackie

Thanks, Pascale

deustp01 commented 5 years ago

The two families have quite distinct structures and accomplish the transport and energy consumption by quite different mechanisms - Alberts, Molecular Biology of the Cell, 5th ed, pp 654-5, 659-666.

pgaudet commented 5 years ago

I know the mechanism is different, but I thought that was not in the scope of GO. That being said, I dont mind having them in. (especially, they are all correctly annotated!)

The next question would then be, can we go ahead and call them 'P-type ATPase transporters'? rather than 'ATPase activity, coupled to transmembrane movement of ions, phosphorylative mechanism' (or is this clear what it is ?

Next-next question would be to have a similar grouping term for ABC transporters - right now they are all directly under the parent

image

I think there was a decision in the past NOT to use 'ABC transporter', but could we use 'ABC-type transporter activity?

Thanks, Pascale

pgaudet commented 5 years ago

Looks like Rhea merged ABC transporters and P-types, see for eg https://www.rhea-db.org/reaction.xhtml?id=14633

Is this right @amorgat ?

deustp01 commented 5 years ago

@amorgat And could you say a bit about the rationale for the merge, if so? The Alberts Molecular Biology of the Cell textbook makes a big point about the distinct molecular properties of the ABC and P-type transporters, so I'm curious as it's not an area I really know (but we should all annotate it consistently). Thanks

amorgat commented 5 years ago

@deustp01 and @pgaudet. To give you a proper feed-back, I need to wait Kristian Axelsen (@kaxelsen) who is in vacation this week. Kristian will be back the June 3. Sorry for the delayed answer!

pgaudet commented 5 years ago

Just for the record: This is what the branch looked like before I stated the work on this. ABC transporters are named 'ATPase-coupled x transporter' or 'x ABC transporter activity' and P-type usually 'x-transporting ATPase' (although some ABC transporter are also called like that). The only way to know is to look at synonyms.

So I will create two grouping terms 'ABC-type transmembrane transporter activity' and 'P-type transmembrane transporter activity', and rename all children with this pattern.

(Note that I don't intend to do this for flippases and floppases, although some are ABC family and some are P-type 4 family).

image

OK ?

Thanks, Pascale

ValWood commented 5 years ago

It does seem strange to use protein family names to group. If we need to keep them to distinguish the mechanism, it somehow feels more correct to have the mechanism in the term name (also prevent other such 'family' terms).

pgaudet commented 5 years ago

I agree, but the alternative is not pretty.

Both P-type ATPases and ABC transporter act by inducing a conformational change (see https://en.wikipedia.org/wiki/P-type_ATPase#Mechanism and https://en.wikipedia.org/wiki/ATP-binding_cassette_transporter#Mechanism_of_transport)

Two differences:

Adding this information in the term label seems tricky! Doyou have suggestions?

ValWood commented 5 years ago

OK, agreed.

dsiegele commented 5 years ago

I don't know if this is helpful, but I often refer to the Transporter Classification Database (www.tcdb.org) when I have questions about transporters. The TCDB classifies membrane transporter systems using both functional and phylogenetic information.

TCDB classifies ATP-binding cassette transporters and P-type ATPase transporters into two superfamilies. http://www.tcdb.org/search/result.php?tc=3.A.1 The ATP-binding Cassette (ABC) Superfamily http://www.tcdb.org/search/result.php?tc=3.A.3 The P-type ATPase (P-ATPase) Superfamily

The subclasses of GO:0042626 ATPase-coupled transmembrane transporter activity seem to be based on the substrate being transported. The majority of P-type ATPase transporters in prokaryotes and eukaryotes catalyze cation uptake and/or efflux driven by ATP hydrolysis, but some have been shown to transport lipids, for example PMID:21454556. Would there need to be a separate branches for 'P-type ATPase transporter activity, coupled to movement of ions' and 'P-type ATPase transporter activity, coupled to movement of lipids'?

The ABC-type transporters catalyze uptake and/or efflux of a variety of substrates, including cations, anions, sugars, amino acids, glycolipids, siderophores, etc.

kaxelsen commented 5 years ago

@deustp0 Rhea only covers chemical transformations, it does not take the protein performing this transformation into account. So in line with this, the mechanism is not taken into account either. This means that when a P-type ATPase and and an ABC ATPase performs the same overall reaction, it will be covered by the same Rhea ID.

deustp01 commented 5 years ago

From @kaxelsen 's comment, all peptidases should map to a single Rhea reaction in which a molecule of water is incorporated lysing a peptide bond, but that is consistent with maintaining multiple GO MF terms that classify these peptidases further by location of the cleavage site (endo- and exopeptidases) and also by cleavage mechanism (serine peptidases, metallopeptidases, ...), so for molecular function terms (not process terms, as above) the fact that more than one GO term maps to a single Rhea reaction family doesn't look like an obstacle. @pgaudet from @dsiegele 's comment, it looks like there are compact, widely-used names for the two functional kinds of ATP-consuming transporters, so why not use them? For definition purposes, if I understand right, in both cases a conformational change in the transporter causes a ligand binding site initially exposed on one side of a membrane to become exposed on the other side, so that ligand bound on the first side can be discharged on the second. In the case of P type transporters, phosphorylation of the transporter at the cost of an ATP drives the conformational change. In the case of ABC transporters, ATP binding drives the change and unbinding requires ATP hydrolysis so again tyransport happens at the cost of ATP.

pgaudet commented 5 years ago

OK we'll keep both P-type and ABC-type transporters. They will link to the same RHEA ID.

Thanks, Pascale

ukemi commented 5 years ago

If they link to the same Rhea ID, they will have the same definition.

pgaudet commented 5 years ago

you mean textual definitions?

ukemi commented 5 years ago

The plan is to use the Rhea reactions to define GO-MFs, so if two different GO terms point to the same Rhea identifier, they will become equivalent terms.

pgaudet commented 5 years ago

OK - is this what we want to do ? That means GO cannot have different guidelines from Rhea for deciding when a catalytic activity exists - is this what was decided?

Alternatively - we could allow 1:n mappings between Rhea and GO terms and only use Rhea to populate our definitions for 1:1 cases?

deustp01 commented 5 years ago

Existence of catalytic activity - I'm confused - thought Rhea annotated descriptions of the chemical changes that happen in a reaction, not whether those changes are spontaneous or accelerated by a gene product acting as a catalyst.

If it is GO policy that if a Rhea reaction (family) has been annotated then there shall be exactly one GO MF term for enabling that reaction, then all Rhea:GO mappings must be one:one and all MF terms involving chemical reactions map to one Rhea reaction (family) either because it already exists or because GO people work with Rhea people to get the Rhea reaction (family) created.

That logic all looks straightforward, and now that David mentions it, yes, that was a decision from the December Geneva meeting.

Thinking about it now, I wonder if we want it uniformly - do we want all the distinct ways that ATP consumption can be coupled to transport against a concentration gradient, or all the ways that a peptide bond can be hydrolyzed to collapse into single MF terms? That, however, is a question of biologist usability (and maybe of slim uniformity, but that's not my headache), not of logic.

pgaudet commented 5 years ago

Do we want to revisit this decision ? If not I will merge....

pgaudet commented 5 years ago

A few edits I can already do:

GOID Term label Number of direct annotations (non-IEA, non-IBA, non-ISO) Family EC MetaCyc Rhea Other mapping Notes Action
GO:0102022 ATPase-coupled L-arginine transmembrane transporter activity' 4 ABC-type EC:7.4.2.1 ABC-4-RXN 29879   All annot are ISS TIGR Merge into 'A GO:0061459 L-arginine transmembrane transporter activity
GO:0015610 glycerol phosphate-importing ATPase activity' 0   NA NA NA TC:3.A.1 (general ABC transporter) Merge into GO:0015430 glycerol-3-phosphate-transporting ATPase activity
GO:0015609 maltooligosaccharide-importing ATPase activity' 0   NA NA NA TC:3.A.1 (general ABC transporter) Merge into GO:0015422 ATPase-coupled oligosaccharide transmembrane transporter activity
GO:0051471 ectoine transmembrane transporter activity' 0 ABC-type NA NA NA NA   Merge into GO:0033286 ATPase-coupled ectoine transmembrane transporter activity
GO:0033287 hydroxyectoine transmembrane transporter activity' 0 ABC-type NA NA NA NA   Merge into  GO:0033288 ATPase-coupled hydroxyectoine transmembrane transporter activity
pgaudet commented 5 years ago

Create 2 subfamilies, ABC transporter and P-types: