Include shapes for models/graphs

[cmungall]

We should include shapes for the model itself.

I think we should make all assumptions explicit. E.g. should every model have a title? a status?

Should we have separate shapes for a curated model vs the named graphs we use to cache inferences (cc @balhoff @dougli1sqrd )?

here is a model chosen at random

http:https://model.geneontology.org/5b318d0900000583,http:https://www.geneontology.org/formats/oboInOwl#id,gomodel:5b318d0900000583
http:https://model.geneontology.org/5b318d0900000583,http:https://model.geneontology.org/graphType,http:https://model.geneontology.org/noctuaCam
http:https://model.geneontology.org/5b318d0900000583,lego:modelstate,production
http:https://model.geneontology.org/5b318d0900000583,http:https://purl.org/pav/providedBy,http:https://www.wormbase.org
http:https://model.geneontology.org/5b318d0900000583,owl:versionIRI,http:https://model.geneontology.org/5b318d0900000583
http:https://model.geneontology.org/5b318d0900000583,https://w3id.org/biolink/vocab/in_taxon,obo:NCBITaxon_6239
http:https://model.geneontology.org/5b318d0900000583,rdf:type,owl:Ontology
http:https://model.geneontology.org/5b318d0900000583,owl:imports,obo:go/extensions/go-lego.owl
http:https://model.geneontology.org/5b318d0900000583,dc:title,C. elegans PERK-mediated unfolded protein response
http:https://model.geneontology.org/5b318d0900000583,dc:contributor,http:https://orcid.org/0000-0002-1706-4196
http:https://model.geneontology.org/5b318d0900000583,dc:contributor,http:https://orcid.org/0000-0002-3013-9906
http:https://model.geneontology.org/5b318d0900000583,dc:date,2019-11-26

here is it's corresponding inferred graph:

http:https://model.geneontology.org/5b318d0900000583_inferred,http:https://model.geneontology.org/graphType,http:https://model.geneontology.org/noctuaCam
http:https://model.geneontology.org/5b318d0900000583_inferred,prov:wasDerivedFrom,http:https://model.geneontology.org/5b318d0900000583

todo: compare with a reactome-derived graph @goodb

Do we have a more explicit way to tell if a graph is inferred other than checking if a derivedFrom triple is there? Is this a guarantee, e.g. might we have derivedFrom on the main models in future (e.g. derived from a template?)

[goodb]

@cmungall we do have the start of this in the schema now. e.g. it specifies that models should have exactly one title

<GoCamModel> {
  a [owl:Ontology] + ;
  contributor: xsd:string +; #TODO would be better as an IRI
  date: xsd:string {1}; #TODO can we make this an xsd:date?
  provided_by: xsd:string +; #TODO would be better as an IRI
  rdfs:comment xsd:string *;
  modelstate: xsd:string {1}; #TODO would be better as an IRI
  in_taxon: . *;
  title: xsd:string {1};
  imports: . *;
  oboinowlid: . *; #TODO not sure if we really want this?
  owl:versionIRI . *;
}

[cmungall]

ah cool, was looking at an older version so if we run this over the public triplestore all the inferred models will be marked invalid (same if we adopt a caching reasoner strategy for the curation store) it's also confusing if we use shex as documentation for end-user developers any ideas how to handle this? Two separate GoCamModel classes?

…

On Mon, Nov 16, 2020 at 4:00 PM goodb ***@***.***> wrote: @cmungall <https://github.com/cmungall> we do have the start of this in the schema now. e.g. it specifies that models should have exactly one title <GoCamModel> { a [owl:Ontology] + ; contributor: xsd:string +; #TODO would be better as an IRI date: xsd:string {1}; #TODO can we make this an xsd:date? provided_by: xsd:string +; #TODO would be better as an IRI rdfs:comment xsd:string *; modelstate: xsd:string {1}; #TODO would be better as an IRI in_taxon: . *; title: xsd:string {1}; imports: . *; oboinowlid: . *; #TODO not sure if we really want this? owl:versionIRI . *; } — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#252 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAMMON27OHAI37FCVJ4W6DSQG4JLANCNFSM4TX273TA> .

[cmungall]

current count of model properties (on public store):

N	property
52	rdfs:label
66	lego:json-model
130	owl:versionIRI
301	http:https://www.geneontology.org/formats/oboInOwl#id
2237	rdfs:comment
2932	https://w3id.org/biolink/vocab/in_taxon
2960	lego:modelstate
2960	dc:date
2961	dc:title
2970	http:https://purl.org/pav/providedBy
2979	owl:imports
3013	http:https://model.geneontology.org/graphType
3013	rdf:type
5541	dc:contributor

• there are some above not accounted for in the schema (e.g graphType) - ideally schema would be closed
• I thought json-model was gone?
• presumably owl:imports will soon be gone
• Can we delete all:
  • oio:id
  • rdfs:label (why have this and title?)