Roles doing stuff where entities should be?

[dosumis]

It seems odd to have roles rather than entities inhibiting, activating etc.

e.g. '4-aminobutyrate transaminase inhibitor' (a CHEBI ROLE) EquivalentTO 'biological role'
and (inhibitor_of some '4-aminobutyrate transaminase activity')

This could work:

'chemical entity' that has_role some 4-aminobutyrate transaminase inhibitor' EquivalentTo inhibitor_of some '4-aminobutyrate transaminase activity'

But perhaps harder to maintain, in which case GO(CHE) style bridging classes might be a good idea.

Alternatively, in the spirit of the BFO conception of this, we might say

4-aminobutyrate transaminase inhibitor (role) EquivalentTo: realised_by some* (inhibition_of some 4-aminobutyrate transaminase activity')

(*only ??? - I don't think this is necessary if roles can be bourne without being realised)

Or combining with a typical GO way of doing things:

4-aminobutyrate transaminase inhibitor (role) EquivalentTo: realised_by some 'negative regulation of 4-aminobutyrate transaminase activity')

Perhaps some use case queries would help clarify the preferred approach.

[cmungall]

• = I think it formally needs to be "only". Otherwise every instance of the chemical entails the existence of a process. I don't think this is problematic for various formal and pragmatic reasons but others may object.

I'd rather keep it EL for the first pass, so at some stage we should auto-transform all of these to the pattern you suggest.

• also inhibition -> regulation (current snapshot is partly a reflection of the semi-automated methods used to construct)