Axiomization of molecular functions

[cthoyt]

I've been pushing to include more relations into the Relation Ontology that correspond to various post-translational modifications (e.g., phosphorylates, ubiquitinates, myrisotlyates).

I began curating a list of relationships, their GO molecular functions that need to get "rolified" as well as the chemicals corresponding to the groups that get put on / taken off by genes with those functions. For example, the myristoylates relation corresponds to myristoyltransferase activity (GO:0019107) which puts on a tetradecanoyl group (CHEBI:25456).

Curation is here: https://github.com/cthoyt/obo-relations-interactions/blob/main/relations.tsv (CC0 license). Would be great if you could reuse some of it!

[cmungall]

Hi @cthoyt - I think it should be easy to add these as simple has-participant relationships. But note this doesn't designate the role of the participant. These would be subClassOf axioms, not equivalent class (otherwise kinase and dephosphatase would collapse). Did you have in mind a more granular representation where GO would capture the distinction between addition to and removal front the target?

[cthoyt]

@cmungall sorry for the late reply. I don't think there is a need to represent the more granular distinction between addition and removal on a first pass. I don't think GO is really trying to capture this information so maybe it's not appropriate at all

I also recently found that UniProt has done a lot of interesting curation linking PTMs themselves to the residues where they happen (ref; https://www.uniprot.org/news/2019/07/31/release) in their https://www.uniprot.org/docs/ptmlist

[alanbridge]

Hi Charles,

the PTM mapping to ChEBI is led by Elisabeth Coudert at Swiss-Prot (who is also leading a project to convert all existing text annotations of small molecule ligands to ChEBI too - features like BINDING, METAL and others will be "ChEBIfied" this year, the overall goal being to put all small molecule annotations into the ChEBI namespace).

The PTM-ChEBI mapping will make PTM data in UniProt consistent and interoperable with the descriptions of reactions that create/remove PTMs from Rhea. Don't hesitate to drop Elisabeth a line if you need more info.

https://www.sib.swiss/directory?cn=elisabeth%20coudert

All the best, Alan

[pgaudet]

I will close this since there is no action needed for GO.

[cthoyt]

@pgaudet I think there might have been a misunderstanding - there's still an action item for GO to add has-participant relationships. My comment earlier today was only that we shouldn't go overboard with creating a new formalisms for more granular information.

[pgaudet]

Sorry, just to clarify: Do you propose to add for example 'GO:0019107 myristoyltransferase activity' has participant CHEBI:25456 tetradecanoyl group?

[cthoyt]

Short answer, yes - something like that.

Well the formalism itself is up to GO, but I have curated pairs of GO molecular functions and ChEBI molecular groups that could be formalized this way. @cmungall recommended I make the GO curators aware of this so they could consider adding some kind of relation (but again it would be up to you which one you think is most appropriate).

[pgaudet]

OK, let's clarify this with @cmungall

I thought these were supposed to come from RHEA.

Thanks for the explanation!

[alanbridge]

Hi Pascale,

you won't normally get such high level grouping terms from Rhea unless these are really needed for annotation (this GO term says nothing at all about the substrate, nor about the myristoyl donor) - so there will be many higher level nodes in the GO with no Rhea.

All the best, Alan