enanomapper / slimmer Goto Github PK

I got the following error message when trying to convert the NPO ontology into OBO format, pointing at the need for a feature to merge rdfs:comment labels in slimmer.

OBO STRUCTURE ERROR Ontology does not conform to OBO structure rules:
multiple comment tags not allowed. in frame:Frame(Disposition id( Disposition)name( disposition)comment(

Possibly an OWLAPI issue, which does the serialization as RDF/XML.

See this upstream issue: owlcs/owlapi#439

Janna, are there alternative download locations? Jenkins reports:

--2015-09-08 06:43:09--  (try:11)  http://livercancer.imbi.uni-heidelberg.de/ccont.owl
Connecting to livercancer.imbi.uni-heidelberg.de (livercancer.imbi.uni
    -heidelberg.de)|129.206.228.156|:80... failed: Connection timed out.
Retrying.

There seems to be something odd going on with the slimmer.

In CHEMINF, despite clearly specifying the same parent class for 'molecular entity quality', 'chemical substance quality' and 'surface area' which I newly added to the cheminf.iris for export, only one of the three is getting the correct parent allocated, while the other two are assigned no parent class.

In PATO, newly assigned classes for export, e.g. size, are not being included in the resulting slim file, despite the build executing on commit as usual.

Possible via a local void file?

Something like:

+D(http://new.super/Class):http://www.ifomis.org/bfo/1.1/snap#DependentContinuant

Currently the slimmer only accepts and "slims" classes. We need to add relationships, which can also be duplicated in much the same fashion as classes are.

The following config line does not seem to be doing what it should do:

+D(http://purl.bioontology.org/ontology/npo#NPO_707):http://purl.obolibrary.org/obo/CHEBI_50828	silicon dioxide nanoparticle 

See enanomapper/ontologies#106 (comment)

Why were the terms "skin sensitization" and "skin irritation" not put under "dermal toxicity" but made a separate class. These terms were manually added to the eNM ontology so please can you explain why this decision was made @jannahastings ?

This includes:

• version it was derived from
• date
• Jenkins job(?) or just "eNanoMapper"
• ...

I have added the .props and .iris

There seems to be a change in the subclasses present in the headclass of "toxicity" which is now under the superclass of "information content entity". Before this was under "disposition" so something must have changed in composing the eNM ontology. NOTE: under the toxicity class there are some classes which have manually been added to the eNM ontology.

Testing adding relationship IRIs for BAO, although the build shows as successful the result is strange (no labels, relationships are still there) and there is an error report in the workspace.

Error report: https://jenm.bigcat.maastrichtuniversity.nl/job/BAO/ws/hs_err_pid2187.log/*view*/

org.semanticweb.owlapi.model.OWLRuntimeException: 1 pending annotations should have been used by now. [Annotation(http://purl.obolibrary.org/obo/IAO_0010000 http://purl.obolibrary.org/obo/BFO_0010034)]
at org.semanticweb.owlapi.rdf.rdfxml.parser.OWLRDFConsumer.setPendingAnnotations(OWLRDFConsumer.java:567)
at org.semanticweb.owlapi.rdf.rdfxml.parser.TripleHandlers$TypeAxiomHandler.handleTriple(TripleHandlers.java:2752)
at org.semanticweb.owlapi.rdf.rdfxml.parser.TripleHandlers$HandlerAccessor.applyAnnotations(TripleHandlers.java:170)
at org.semanticweb.owlapi.rdf.rdfxml.parser.TripleHandlers$HandlerAccessor$3.handleResourceTriple(TripleHandlers.java:199)
at org.semanticweb.owlapi.rdf.rdfxml.parser.OWLRDFConsumer.iterateResourceTriples(OWLRDFConsumer.java:2484)
at org.semanticweb.owlapi.rdf.rdfxml.parser.TripleHandlers$HandlerAccessor.consumeAnnotatedAxioms(TripleHandlers.java:194)
at org.semanticweb.owlapi.rdf.rdfxml.parser.TripleHandlers$HandlerAccessor.mopUp(TripleHandlers.java:377)
at org.semanticweb.owlapi.rdf.rdfxml.parser.OWLRDFConsumer.endModel(OWLRDFConsumer.java:1494)
at org.semanticweb.owlapi.rdf.rdfxml.parser.RDFParser.parse(RDFParser.java:134)
at org.semanticweb.owlapi.rdf.rdfxml.parser.RDFXMLParser.parse(RDFXMLParser.java:81)
at uk.ac.manchester.cs.owl.owlapi.ParsableOWLOntologyFactory.loadOWLOntology(ParsableOWLOntologyFactory.java:159)
at uk.ac.manchester.cs.owl.owlapi.OWLOntologyManagerImpl.loadOntology(OWLOntologyManagerImpl.java:958)
at uk.ac.manchester.cs.owl.owlapi.OWLOntologyManagerImpl.loadOntologyFromOntologyDocument(OWLOntologyManagerImpl.java:896)
at uk.ac.manchester.cs.owl.owlapi.OWLOntologyManagerImpl.loadOntologyFromOntologyDocument(OWLOntologyManagerImpl.java:918)
at com.github.enanomapper.Slimmer.(Slimmer.java:62)
at com.github.enanomapper.Slimmer.(Slimmer.java:48)
at com.github.enanomapper.Slimmer.main(Slimmer.java:119)

So that they are exposed via purl.enanomapper.org.

There might be a bug with how the parent class (specified in ()) is assigned in the case of cheminf. For cheminf I included only one class with descendents -- chemical descriptor -- and assigned it the parent IAO_0000030 i.e. information content entity. However, in the resulting SLIM the class has no specified parent. I suspect this is because the IAO class that I want to delegate as parent IS actually present in the source version of cheminf, so it gets removed along with the rest of the imported IAO. However, despite the fact that it is present in cheminf I don't want to import it from there as we are importing direct from IAO.

After the slimmer has executed, the resulting file (xxx-slim.owl) should have at least one class in it. Do we agree and if yes should we add that as an automated test?

update CITATION.cff to 1.2.0

This will allow two Jenkins jobs, one for testing the loading, with the intention to turn red and complain on Slack when that happens, and a second Jenkins jobs for the labels and descriptions.

• upgrade to JUnit 5
• add tags

Taking input from the .props files.

The latest version of the NPO slim file is not usable, as it generates an error in Protege / other OWL parsers. According to the error message from Protege, the problem is with an unknown prefix “ncicp” in the XML. This appears in the syntax used internally in the NPO for their definitions, e.g.

<npo:FULL_SYN
rdf:datatype="http://www.w3.org/2001/XMLSchema#string">&lt;ncicp:ComplexTerm
xmlns:ncicp="http://ncicb.nci.nih.gov/xml/owl/EVS/ComplexProperties.xsd#&quot;&gt;&lt;ncicp:term-name&gt;cylindrical
shape</ncicp:term-name><ncicp:term-group>PT</ncicp:term-group><ncicp:term-source>NCI</ncicp:term-source></ncicp:ComplexTerm>/npo:FULL_SYN

In previous versions of the slimmer, the output file produced for the NPO slim did not include the escaping of XML characters in the embedded definition, e.g. it used to include ncicp:ComplexDefinition whereas now it includes <ncicp:ComplexDefinition> This change must reflect an underlying change in the slimming software since nothing has changed in the underlying NPO OWL file.

The OBI slim file includes the class BFO_0000003 (without metadata) which is not included in the instructions. The only place that the class is used in the slim file is in the range of the preceded_by and precedes relations. As this class is not included in the instructions I would not expect it to be included in the resulting slim file.

In the owl:Ontology annotation.

@jannahastings , what license shall we use? I'm happy with the code under the MIT license. What about the config files? Also MIT or do you want to specify something else?

Update pom.xml to have the latest version of owlapi-distribution

<dependency>
		<groupId>net.sourceforge.owlapi</groupId>
		<artifactId>owlapi-distribution</artifactId>
		<version>4.3.2</version>
	</dependency>

I did a copy and updated to version 5.1.11 but had issues compiling. Changes at the java code might be required.

Ref.:
https://mvnrepository.com/artifact/net.sourceforge.owlapi/owlapi-distribution