NTR: Add 3 pharmacological relations for small molecules

[cthoyt]

Closes #371

Motivation

For ChIRO and the Chemical Roles Graph (https://github.com/chemical-roles/chemical-roles), we need more detailed relationships to describe causal interactions between chemicals (material entities) and receptors (material entity).

Existing relationships (i.e., directly negatively regulates activity of and directly positively regulates activity of) are not detailed enough.

See similar issue about drug-target relationships in https://github.com/oborel/obo-relations/issues/144.

What this PR does

This PR creates 3 new properties:

LUID	Name	Parent	Example
0018027	is agonist of	directly positively regulates activity of	mibolerone (CHEBI:34849) is agonist of androgen receptor (PR:P10275)
0018028	is inverse agonist of	directly negatively regulates activity of	pimavanserin (CHEBI:133017) is inverse agonist of HTR2A (PR:P28223)
0018029	is antagonist of	directly negatively regulates activity of	tretinoin (CHEBI:15367) is antagonist of Nuclear receptor ROR-beta (PR:Q92753)

Big thanks to Jie, Alexander, Darren, Bill, and everyone else at the 2022-11-01 RO meeting who gave feedback and helped improve this PR.

[sierra-moxon]

Would this also set precedence for adding other, more specific mechanism properties? e.g. mechanisms in biolink

[cthoyt]

I think it would be totally fantastic to have biolink-like properties standardized in RO as well!

[deepakunni3]

+1 to standardizing Biolink predicates/properties in RO 🚀

[wdduncan]

Notes for RO meeting:

• add domain and ranges
• perhaps we need versions that are molecule to molecule, and molecule to process
• place properties under more general super-properties that have correct domains and ranges

[balhoff]

@cthoyt maybe we could create some test GO-CAMs and then decide which nodes you would like these relations to go between. It also might help us to define any additional rules that would be useful.

[cthoyt]

@wdduncan in the last few commits, I moved these relations under the directly negatively regulates activity of and directly positively regulates activity of, which have clear domain and range annotations both being material entities. The description of these properties directly addresses the comments made by @addiehl to better notate that the existence of a relation mibolerone (CHEBI:34849) is agonist of androgen receptor (PR:P10275) implies that there is some activity of PR:P10275 that is getting modulated.

I found the example usage annotation and added some explicit examples for each both in the PR description and in the ontology.

I added seeAlso links to related Wikidata properties.

I improved the descriptions for each as well.

[cthoyt]

@balhoff here's a first stab at the relations that might be in a GO CAM

subject name	subject curie	relation name	relation curie	object name	object curie
tretinoin	CHEBI:15367	is antagonist of	RO:0018029	Nuclear receptor ROR-beta	PR:Q92753
Nuclear receptor ROR-beta	PR:Q92753	enables	RO:0002327	activity nuclear receptor activity	GO:0004879
tretinoin	CHEBI:15367	is small molecule inhibitor of	RO:0012006	activity nuclear receptor activity	GO:0004879

(updated based on comments from Jim)

[wdduncan]

Thanks @cthoyt ! This area is out of my domain of expertise. So, I am going to defer to those more knowledge than me about whether this PR is read to be merged :)

[balhoff]

@balhoff here's a first stab at the relations that might be in a GO CAM

@cthoyt great, thanks! For your 'has activity' cell, you would use enables. Should there be a way to infer 'is antagonist of'? For example:

'is small molecule inhibitor of' o 'enabled by' -> 'is antagonist of'

Or is there not enough information there to distinguish 'is antagonist of' from 'is inverse agonist of'?

Similar question for is small molecule activator of and 'is agonist of'.

[cthoyt]

@balhoff thank you for the feedback! I don't think we can make this super general as the difference between antagonism and inverse agonism is based on the way that the chemical binds to the receptor. Also, not all 'is small molecule inhibitor of' o 'enabled by' should be 'is antagonist of'. This should only be true if only if we restrict the range of the enabled by to be receptor activity (GO:0004872) and we say something specific about the mechanism of bindin (seems a bit complicated for this first pass)

[sierra-moxon]

In Biolink we define antagonism as "A causal mechanism in which the effector binds to a receptor and prevents activation by an agonist through competing for the binding site."

And inverse agonism in Biolink is "A causal mechanism in which the effector binds to the same receptor-binding site as an agonist and antagonizes its effects, often exerting the opposite effect of the agonist by suppressing spontaneous receptor signaling."

Do these definitions reflect your understanding of the difference @cthoyt?

[cthoyt]

Yeah that’s basically what I wrote, but we can also update the definitions if you think that would be more clear

[wdduncan]

@balhoff @sierra-moxon @addiehl
Am I correct in assuming that you all have approved this PR, and it is ready to be merged?