#99 Alignment with ChEBI

[David Osumi-Sutherland]

The links between ChEBI and Pro appear to be less specific than they could be:

[Term]
id: PR:000018263
name: amino acid chain
def: "A molecular entity that is a polymer of amino acids linked by peptide bonds." [PRO:DAN]
synonym: "peptide" NARROW []
synonym: "polypeptide" NARROW []
is_a: CHEBI:23367 ! molecular entity

More specific ChEBI classifications for this and/or protein would provide useful chemical context. The best candidate for a more specific term to choose from Chebi is
http://www.ebi.ac.uk/chebi/chebiOntology.do?chebiId=CHEBI:16541

This gives the most chemical context.

(For example of why this is useful, see: https://github.com/geneontology/go-ontology/issues/11953 )

[Darren Natale]

Hi David,

Thanks for the input. I can see why you make the claim you make; however, the relationship between the two terms "amino acid chain" and "protein polypeptide chain" is actually the reverse of what you say. Key are the two restrictions built into the latter: (1) that it is synthesized at the ribosome (explicitly stated); and (2) that it is more than 10 amino acids long (inherited from "peptide"). Our "amino acid chain" is restricted only by the need for at least two peptide-bond-attached amino acids. This include those made by ribosomes and those made by other means such as nonribosomal peptide synthetases.

Best regards,
Darren

[David Osumi-Sutherland]

Hi Darren,

I take your point on the specific suggestion, but are you sure you don't want some kind of link back from protein to ChEBI with more detail than a ChEBI root term? This could provide a rich chemical classification that Pro, for completely understandable reasons of scope, lacks:

http://www.ebi.ac.uk/chebi/chebiOntology.do?chebiId=CHEBI:16541&treeView=true#vizualisation

Looking at your objections:

Key are the two restrictions built into the latter: (1) that it is synthesized at the ribosome (explicitly stated); and (2) that it is more than 10 amino acids long (inherited from "peptide")

The 'synthesised at the ribosome clause seems to be in Pro:protein and Chebi:protein polypeptide chain:

Pro:Protein: "An amino acid chain that is produced de novo by ribosome-mediated translation of a genetically-encoded mRNA. [PRO:DAN, PRO:WCB]
ChEBI: protein polypeptide chain: "A naturally occurring polypeptide synthesized at the ribosome."

The amino acid chain length looks like it is still a problem though still be a problem. Could potentially co-ordainte some compromise with ChEBI, but perhaps a better option would be to classify PRO amino-acid chain as a type of ChEBI peptide (CHEBI:1667)

Looks correct to me and has brings in all the same chemical classification as my previous suggestion.

Cheers,
David

[Darren Natale]

Agreed, it does look correct, at least based on definition. We discussed the suggestion at PRO, and will possibly re-configure PRO: amino acid chain under CHEBI: peptide. We do, however, have a few concerns. One is the history behind the label "peptide" which has led us to avoid it's use (that's why we have "amino acid chain"). The label (historically) comes with several implications: that the entity comprises amino acid residues only (that is, X=OH in the ChEBI definition), and that there is an upper limit on length (10, 20, 30, or 50 amino acid residues, depending on whom you ask). Because the label itself has so much built-in preconceived notions, we've requested that the ChEBI team review its use. We've suggested "amino carboxylic acid chain" instead.

[David Osumi-Sutherland]

Hi Darren,

I can see your problem with the ChEBI nomenclature. Certainly in my experience the term peptide is only used for short chains, but the placement as a parent class to polypeptide is not consistent with that. But my understanding of the way linking beween foundry ontologies typically works is via considering definitions, not term names. Certainly waiting for nomenclature agreement can considerably slow down the addition of useful mappings. Getting this mapping would be much appreciated by GO.

Anyway, just my 2¢. Thanks for considering,

David

[David Osumi-Sutherland]

Hi again Darren,

One other suggestion to avoid objections from PRO users based on the name: you could add a comment to PRO:'amino acid chain' making it clear that the mapping is justified despite ChEBI use of the term 'peptide' being broader than many would expect.

Thanks again,
David

[David Osumi-Sutherland]

Hi Darren,

When we discussed this at the GOC DC meeing, I believe you said you'd reconsider this. As I pointed out in my last 2 comments - this is really just a terminology issue, not a definitional one. GO badly needs this resolved. See https://github.com/geneontology/go-ontology/issues/12369 for the lastest case where this is an issue for us.

Many thanks,
David

[Darren Natale]

As you probably saw in my response to the issue you linked, we did indeed reconsider. However, the issue turned out to be definitional after all. I have other ideas I'll look into so that we meet your needs.