Request for adding property created from template to OMO

[wdduncan]

IRI

No response

Label

created from template

Definition of the property

An annotation property that provides the name of the template or an IRI of the template used to create a term.

Parent property

No response

What is the range of the property in question?

xsd:string

Examples of use

A contrived example from UBERON:
'lower central secondary incisor tooth' created from template lower_tooth_pattern.yaml.

Motivation to add

Ontology terms can come from a number of sources: imports, ROBOT templates, DOSDP templates, and perhaps other custom build templates. In #60, dc:source and created_by are discussed as possibilities. It would great to get some resolution about this.

cc @matentzn @cmungall

ORCID, ROR or Wikidata identifier of the contributor

https://orcid.org/0000-0001-9625-1899

OMO Checklist

• I believe the property is generally useful beyond my specific ontology needs.
• There is no other property in OMO that covers the same use case.

[bpeters42]

I agree that such a property can be useful; but should the template be identified by a URI / purl? Just the name of the template seems to be a bit weak - I would not know where to find the one from this example.

[wdduncan]

@bpeters42 In the definition I do mention using an IRI.

An annotation property that provides the name of the template or an IRI of the template used to create a term.

IRI would be better. But, for purposes of maintaining an ontology, the name may suffice. If something like this was adopted, it may even be best practice to only have the annotation in the edit file, and then remove those annotations when the release is built.

[cmungall]

@bpeters42 agreed - all our templates have URIs

@wdduncan - this is always generated by the system so there is no reason to use just the name. And we should not strip out important provenance data from the release

@matentzn we had previously discussed using dc:conformsTo for this. In fact we do this for classes auto-inferred to belong to a template (or to instantiate a metaclass, which is the better way to think of it). I think we should either

• have a single property and use reification to annotate provenance
• OR have a hierarchy:
  • dc:conformsTo
    • assertedToConformTo
    • generatedFrom
    • detectedAsConformingTo

[wdduncan]

this is always generated by the system

@cmungall Are you referring to UBERON? The lower teeth are generated from a dosdp template, but I am not seeing any annotated information about the template, just an obscure database_cross_reference to OBOL:automatic.

You may not want to strip them out of the release for UBERON. That is your call. But, I am not so sure I would want to have them in the OHD release. The information is mainly intended for editors (in my case). But, I am willing to think about this more. In any case, how the annotations are maintained is a different subject.

[matentzn]

I regret a bit that we are using http://www.geneontology.org/formats/oboInOwl#source in dosdp tools at the moment.

In Mondo we use dc:conformsTo and a fake http://purl.obolibrary.org/obo/mondo#should_conform_to

The PROV way of doing it would be very verbose, as there is no shortcut that adequately describes this relation. the next best thing is prov:wasDerivedFrom, but this is more to describe the relation of one dataset that was transferred into another.

Maybe

• dc:conformsTo
  • OMO:generatedFrom

Is the right way to go. If we agree, I will add these.

[wdduncan]

These sounds great to me. Thanks @matentzn !

[wdduncan]

@matentzn If you want to mimic PROV, you could name it wasGeneratedFrom. Just a suggestion.

[matentzn]

Mimic in what sense? afaics there is no such thing as wasGeneratedFrom in PROV!

[wdduncan]

I know there is no wasGeneratedFrom in PROV. But there is "wasGeneratedBy". By mimic I meant using the "wasGenerated" part.