search

OntoZoo / ontobee

Ontobee is a linked data server for ontologies. See: http://www.ontobee.org.
27 stars 5 forks source link

Incorrect placement of some classes #67

Open jamesaoverton opened 8 years ago

jamesaoverton commented 8 years ago

In MRO, 'protein complex' has four children:

However a large number of 'X protein complex with Y serotype' and 'X protein complex with Y haplotype' classes are also appearing as children of 'protein complex' in Ontobee.

Take 'BF protein complex with B12 haplotype' as an example, and call it "B12" for short. These classes are defined using equivalentTo axioms. Inferred subClassOf axioms have been added to the OWL file. You can see that B12 is inferred to be subClassOf 'chicken MHC protein complex with haplotype'. B12 also appears under that parent, where it should be. It should not appear under 'protein complex'.

Both OLS beta and AberOWL display the children of 'protein complex' correctly.

The 'protein complex' term is the most important to us, but this problem affects other MRO nodes. It also affects the display of a small number of classes in OBI, but I had not noticed it until MRO was added to Ontobee. I can provide examples from OBI if that would be helpful.

I'd be happy to help debug this, but the code in this repository is years out of date.

yongqunh commented 8 years ago

Hi James,

I downloaded the MRO from the Ontobee MRO link to local, and display it under Protégé. I found that B12 is also listed as a subclass under ‘protein complex’.

Oliver

From: James A. Overton [mailto:notifications@github.com] Sent: Tuesday, January 12, 2016 8:47 AM To: OntoZoo/ontobee Subject: [ontobee] Incorrect placement of some classes (#67)

In MRO, 'protein complex'http://wwwontobeeorg/ontology/MRO?iri=http://purlobolibraryorg/obo/GO_0043234 has four children:

However a large number of 'X protein complex with Y serotype' and 'X protein complex with Y haplotype' classes are also appearing as children of 'protein complex' in Ontobee

Take 'BF protein complex with B12 haplotype'http://wwwontobeeorg/ontology/MRO?iri=http://purlobolibraryorg/obo/MRO_0001468 as an example, and call it "B12" for short These classes are defined using equivalentTo axioms Inferred subClassOf axioms have been added to the OWL file You can see that B12 is inferred to be subClassOf 'chicken MHC protein complex with haplotype'http://wwwontobeeorg/ontology/MRO?iri=http://purlobolibraryorg/obo/MRO_0001518 B12 also appears under that parent, where it should be It should not appear under 'protein complex'

Both OLS betahttp://wwwebiacuk/ols/beta/ontologies/mro/terms?iri=http%3A%2F%2Fpurlobolibraryorg%2Fobo%2FGO_0043234 and AberOWLhttp://aber-owlnet/ontology/MRO#!http%3A%2F%2Fpurlobolibraryorg%2Fobo%2FGO_0043234 display the children of 'protein complex' correctly

The 'protein complex' term is the most important to us, but this problem affects other MRO nodes It also affects the display of a small number of classes in OBI, but I had not noticed it until MRO was added to Ontobee I can provide examples from OBI if that would be helpful

I'd be happy to help debug this, but the code in this repository is years out of date

— Reply to this email directly or view it on GitHubhttps://github.com/OntoZoo/ontobee/issues/67.

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

jamesaoverton commented 8 years ago

In Protege you must run the HermiT reasoner and look at the 'Class hierarchy (inferred)' tab. Then you will see four children for 'protein complex', just as OLS beta and AberOWL show.

e4ong1031 commented 8 years ago

Hi @jamesaoverton, As @yongqunh pointed out, these terms will appear in Protege without HermiT reasoner. And all the ontologies being uploaded to the RDF triple store do not go through any reasoner, so the Ontobee hierarchy does not match the Protege inferred hierarchy.

Also, all these terms share similar equivalent axiom: 'protein complex' and XXXXX. According to ontobee display schema, any term that has such axiom, will be displayed as a child term of 'protein complex'.

jamesaoverton commented 8 years ago

Thank you for looking into this.

Why does the "ontobee display schema" use part of the equivalence axiom?

We build MRO using the best equivalence axioms we can, then let the reasoner infer the hierarchy. We include the inferred subclass axioms in the OWL file. The same method is used by UBERON (@cmungall) and other recent ontologies with rich logical axioms.

By using just one part of the equivalent axioms, Ontobee will incorrectly display any ontology that uses this approach. Other term browsers such as OLS beta and AberOWL display the hierarchy correctly.

Here is another example from UBERON. 'vasculature' has many children in Ontobee. But many of those children, such as 'lung vasculature', should appear under 'vasculature of organ'.

cmungall commented 8 years ago

Yes, OB same for the asserted view in Protege.

OLS gets this right, as well as many other aspects of hierarchy display. See some of my tickets, for example https://github.com/EBISPOT/OLS/issues/10