Best practices for modeling variables in LinkML, and mapping to i-ADOPT

[cmungall]

A subset of uses of slots in LinkML are to record variables and CDEs. LinkML, allows slots to have a unit, with range UnitOfMeasure. This allows for:

slots:
  diss_o2_mass_in_water:
    range: float
    unit:
      ucum: mg.kg-1
      has_quantity_kind: qudt:Concentration

However, this doesn't capture the object of interest (o2) or the "matrix" (water).

Many groups delegate this to an external ontology:

slots:
  diss_o2_mass_in_water:
    range: float
    unit:
      ucum: mg.kg-1
      has_quantity_kind: qudt:Concentration
      exact_mappings:
       - ENVO:nnnn

Or using the ontology class in place of the generic quantity kind.

However, for many use cases it would be useful to inline the additional metadata

Standards such as ISO-11179-3 (which is largely followed) and I-ADOPT provide some way of doing this.

See for example:
https://vocab.nerc.ac.uk/collection/P01/current/DOXMZZXX/?_profile=iop&_mediatype=text/html

iop:hasProperty already corresponds to has_quantity_kind (== qudt:hasQuantityKind), but we don't have a direct equivalent for matrix and object of interest.

• should we add these? to UnitOfMeasure?
• other aspects to add?