a CDK descriptor as OWL

[Egon Willighagen]

Hi all (Leonid, Michel, Nico in particular),

I just made a commit [0] to add the BODO atomCount descriptor
algorithm, and its CDK implementation.

This depends on two new Classes I defined this afternoon [1] to a
molecular descriptor, and a software
module to calculate molecular descriptors:

<owl:Class rdf:about="&resource;CHEMINF_000450">
<rdfs:label>algorithm to calculate a molecular descriptor</rdfs:label>
<owl:equivalentClass>
<owl:Class>
<owl:intersectionOf rdf:parseType="Collection">
<rdf:Description
rdf:about="&resource;CHEMINF_000144"/> <!-- algo to calc a chemical
descriptor -->
<owl:Restriction>
<owl:onProperty
rdf:resource="&obo;OBI_0000299"/> <!-- has specified output -->
<owl:someValuesFrom
rdf:resource="&resource;CHEMINF_000065"/> <!-- molecular entity
descriptor -->
</owl:Restriction>
</owl:intersectionOf>
</owl:Class>
</owl:equivalentClass>
</owl:Class>

<owl:Class rdf:about="&resource;CHEMINF_000451">
<rdfs:label>software module to calculate a molecular
descriptor</rdfs:label>
<owl:equivalentClass>
<owl:Class>
<owl:intersectionOf rdf:parseType="Collection">
<rdf:Description
rdf:about="&resource;CHEMINF_000103"/> <!-- software module to
calculate a chemical descriptor -->
<owl:Restriction>
<owl:onProperty
rdf:resource="&obo;OBI_0000299"/> <!-- has specified output -->
<owl:someValuesFrom
rdf:resource="&resource;CHEMINF_000065"/> <!-- molecular entity
descriptor -->
</owl:Restriction>
</owl:intersectionOf>
</owl:Class>
</owl:equivalentClass>
<rdfs:subClassOf>
<owl:Restriction>
<owl:onProperty
rdf:resource="&resource;CHEMINF_000047"/> <!-- conforms to -->
<owl:someValuesFrom
rdf:resource="&resource;CHEMINF_000450"/> <!-- see above -->
</owl:Restriction>
</rdfs:subClassOf>
<dc:description>A software module to calculate a molecular
descriptor is software module that implements an algorithm which
calculates a descriptor value for a molecular entity.</dc:description>
</owl:Class>

Using these I added an algorithm from the BODO:

<owl:Class rdf:about="&resource;CHEMINF_001000">
<rdfs:label>Atom Counting algorithm</rdfs:label>
<owl:sameAs rdf:about="&chemoinformatics-algorithms;atomCount" />
<owl:equivalentClass>
<owl:Class>
<owl:intersectionOf rdf:parseType="Collection">
<rdf:Description
rdf:about="&resource;CHEMINF_000450"/> <!-- molecular descriptor
algorithm (see above) -->
<owl:Restriction>
<owl:onProperty
rdf:resource="&obo;OBI_0000299"/> <!-- has specified output -->
<owl:someValuesFrom
rdf:resource="&resource;CHEMINF_000263"/> <!-- atom count descriptor
-->
</owl:Restriction>
</owl:intersectionOf>
</owl:Class>
</owl:equivalentClass>
</owl:Class>

and the matching CDK implementation:

<owl:Class rdf:about="&resource;CDK_000001">
<rdfs:label>The Chemistry Development Kit Library</rdfs:label>
<rdfs:subClassOf rdf:resource="&resource;CHEMINF_000342"/>
<!-- software library -->
</owl:Class>

<owl:Class rdf:about="&resource;CDK_000100">
<rdfs:label>org.openscience.cdk.qsar.descriptors.molecular.AtomCountDescriptor</rdfs:label>
<owl:equivalentClass>
<owl:Class>
<owl:intersectionOf rdf:parseType="Collection">
<rdf:Description
rdf:about="&resource;CHEMINF_000451"/> <!-- see above -->
<owl:Restriction>
<owl:onProperty
rdf:resource="&resource;CHEMINF_000047"/> <!-- conforms to -->
<owl:someValuesFrom
rdf:resource="&resource;CHEMINF_001000"/> <!-- see above -->
</owl:Restriction>
</owl:intersectionOf>
</owl:Class>
</owl:equivalentClass>
<ro:part_of rdf:resource="CDK_000001" />
</owl:Class>

(There is no version number of the software library yet...)

Does this make sense?

Egon

0.http://code.google.com/p/semanticchemistry/source/detail?r=170
1.http://code.google.com/p/semanticchemistry/source/detail?r=169

--
Dr E.L. Willighagen
Postdoctoral Research Associate
University of Cambridge
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: http://www.citeulike.org/user/egonw/tag/papers

[Michel Dumontier]

Looks good to me :)

m.

--
Michel Dumontier
Associate Professor of Bioinformatics
Carleton University
http://dumontierlab.com

[Leonid Chepelev]

Hi Egon,

I am sending separately a populated CHEMINF ontology that I created for the CHEMINF paper. Just open cheminf.owl in the latest protege (check your gmail).

Please note that this prototype had a couple of terms that are slightly different from the cheminf.owl on SVN.

This should give you and idea of how we specify descriptors, algorithms, and software.

Cheers,

Leo C.

> Date: Sat, 4 Dec 2010 16:26:11 -0500
> Subject: Re: a CDK descriptor as OWL
> From: michel.d...@gmail.com
> To: cheminf-...@googlegroups.com

[Janna Hastings]

Hi Leo, all,

Egon's version might be perhaps slightly closer to the current version
of the paper draft than is the generated data, since I was updating
the ontology as I updated the paper these past few weeks, but best
would be if we can all agree on the current representation.

Ideally we should converge on just one, maintainable version of the
BODO mapping and the CDK implementations that lives in the files Egon
has created which are referenced from the main cheminf.owl. These
mappings will bring us many new users in the CHEMINF community thanks
to BODO's and CDK's broad acceptance.

Best wishes,
Janna