Hypothetical modeling of specific pharmacological modes of action

[andrewsu]

Naive question here... In the context of Translator, it seems likely that we will eventually want to model some detailed pharmacological modes of action, e.g., "antagonist" vs "inverse agonist". As a strawman to better understand BioLink Model (and in advance of the formal discussion/approval of these decisions), can someone walk me through the way this would be modeled?

I see an existing slot for negatively regulates, entity to entity:

  negatively regulates, entity to entity:
    aliases: ['activity directly negatively regulates activity of']
    is_a: regulates, entity to entity
    mixins:
      - negatively regulates
    local_names:
      translator: negatively regulates
      ro: activity directly negatively regulates activity of
    in_subset:
      - translator_minimal
    slot_uri: RO:0002449
    mappings:
      - SEMMEDDB:INHIBITS

So would antagonism and inverse agonism be modeled as another subclass slot? Would they be direct subclasses of negatively regulates, entity to entity, or would you create some intermediate class like negatively regulates, chemical substance to gene or gene product? Am I in the right ballpark here?

[andrewsu]

... and I just realized I raised a similar question in #282. So I'm rereading the discussion over there as well...

[nlharris]

Do we need to keep both this and #282 open?

[nlharris]

can this be closed if it's redundant with #282?

[andrewsu]

they are for different edge types (gene-disease vs drug-gene). Perhaps if one of them were explained/closed then the other would be clear too...

[sierra-moxon]

@andrewsu - I think your original use case in this thread, is aided by the refactoring of the chemical predicates (we add agonist and inverse agonist, etc as part of this PR). Would your team mind taking a look through that PR with regards to this ticket? I've linked this ticket to that PR so that when it is merged, this ticket will be resolved.

[andrewsu]

This new structure makes perfect sense to me, thanks! so 👍 if this issue gets closed on merge...