equivalentClass/subClassOf issues

[tgbugs]

We have encountered unexpected behavior when dealing with some of our old defined classes.

In the definition of NIFNEURNT:nlx_neuron_nt_090803 here the equivalence statement is sufficient in Protege for NIFNEURNT:nlx_neuron_nt_090803 to be inferred as subClassOf NIFCELL:sao1417703748. However when looking at the scigraph output for a neighbors query with depth of 2 this relation does not appear. Running the query with relationshipType=equivalentClass set returns nothing (I think this is expected behavior?)

This seems to be related to #135 and #42 but I'm not entirely sure if it is the same issue.

[cmungall]

Looks like #135 is the more relevant ticket

What's your configuration - are you using a reasoner?

[tgbugs]

Reasoner is off (havn't gotten nifstd into reasonable state yet). Configuration is essentially as seen here: https://github.com/SciCrunch/NIF-Ontology/blob/master/scigraph/make_yamls.sh

[cmungall]

In that case, this is expected (if undocumented) behavior.

Although in principle we could implement a reasonerless relaxation step,
that is better kept as the preserve of ontology processing and release
tools upstream of scigraph, e.g. robot:
ontodev/robot#7

Even if the ontology is messy, I'd recommend using Elk as part of the SG
load. You can configure SG to drop the disjointness axioms and
domain/range constraints (which will eliminate issues if you have messy
object properties or object property usage). Unless you have really
heinous equivalence axioms, the results should be trustable.

On 6 May 2016, at 18:04, Tom Gillespie wrote:

Reasoner is off (havn't gotten nifstd into reasonable state yet).
Configuration is essentially as seen here:
https://github.com/SciCrunch/NIF-Ontology/blob/master/scigraph/make_yamls.sh

---

You are receiving this because you commented.
Reply to this email directly or view it on GitHub:
#180 (comment)

[jnguyenx]

Additional note: It should not be related to #135 as it is only activated when this configuration type is specified.

[tgbugs]

Adding the reasoner results in the following warning
WARNING: Not reasoning on Ontology(OntologyID(OntologyIRI(<http://ontology.neuinfo.org/NIF/ttl/nif.ttl>))) [Axioms: 0 Logical Axioms: 0] because it is inconsistent. so I will indeed need to clean up NIFSTD wrt reasonability before I can resolve this. If the issues persists after the reasoner can run I will reopen. Closing for now.

[cmungall]

There's an option to remove disjointness and domain/range axioms prior to loading. These are usually the cause, and they don't really buy you much at the SciGraph level

[tgbugs]

Do you happen to have the reference for that option on hand, or know where I can look? (I checked the Elk documentation but nothing jumped out at me).

[jnguyenx]

That's should be it:
https://github.com/monarch-initiative/SciGraph-docker-monarch-data/blob/master/src/main/resources/monarchLoadConfiguration.yaml.tmpl#L19