Add a predicate specifically for structural relationships between molecular structures

[RichardBruskiewich]

Scientific issue motivating this feature request:

• Need to add chemical structural relationship mappings from relevant ontologies/controlled vocabularies to Biolink Model edge_labelpredicate space.

Possible solution:

Add a new Biolink Model predicate?

  structurally related to:
    is_a: related to
    description: holds between two molecular entities which have highly correlated primary chemical structures
    domain: molecular entity
    range: molecular entity
    exact_mappings:
      -
    broad_mappings:
      -
    narrow_mappings:
      # sample narrow mappings borrowed from S. Ramsey's recent curation of RTX... There may be others?
      - CHEBI:has_parent_hydride
      - CHEBI:is_conjugate_acid_of
      - CHEBI:is_conjugate_base_of
      - CHEBI:is_enantiomer_of
      - CHEBI:is_tautomer_of

Origin of this request:

The need for this change suggested by ongoing data modeling discussions about representing chemical similarities in the Biolink and some attempted mappings of chemical predicates (e.g. from CHEBI) to the model.

Relevant members for discussion @cmungall, @cbizon, Steve Ramsay, Vlado Dancik

[cbizon]

Would this be a child of a proposed similar_to predicate? (see https://github.com/biolink/biolink-model/pull/404)

[RichardBruskiewich]

Hi @cbizon, sure.. that sounds reasonable. Perhaps the name of the slot could also be tweaked to read structurally similar to? Do we need to add any reference to chemistry in the name, i.e. chemical structure similar to to be more precise? Or molecular structure similar to (to better align with the associated domain and range category names)?

[vdancik]

add label: chemical

[RichardBruskiewich]

Hi @vdancik, what did you specifically have in mind? Are you thinking of biolink:chemical_structure_similar_to? I just realised that after previously revisiting this issue, I followed @cbizon's link over to https://github.com/biolink/biolink-model/pull/404 which added "biolink:chemically_similar_to", which was just merged.

That said, there was mention in that PR of an additional refinement of chemistry semantics:

structurally similar to  # could also simply be 'chemical structure similar to'?
    is_a chemically similar to
          is_a similar to
              is_a related to

Vlado, do you think that any further chemistry refinement of biolink terms would be useful, e.g. a term chemical structure similar to plus other related chemistry terms, e.g. predicates like chemical ion form of, chemical radical of or chemical salt of (perhaps others)?

[vdancik]

@RichardBruskiewich I had in mind add github label "chemical" to this issue.

[nlharris]

Is this covered by the chemical entity refactoring that @vdancik is working on? Seems like it should not be assigned to @cbizon?

[sierra-moxon]

@vdancik @RichardBruskiewich - I'm going to close this issue for now as it appears that Richard and Chris B handled at least the basic use case in #404. Please do reopen if I got this wrong! :)