[OTDev] validation search functions with model

Nina Jeliazkova jeliazkova.nina at gmail.com
Wed Aug 24 14:32:15 CEST 2011


On 24 August 2011 15:23, Christoph Helma <helma at in-silico.ch> wrote:

> On Wed, 24 Aug 2011 14:28:40 +0300, Nina Jeliazkova <
> jeliazkova.nina at gmail.com> wrote:
> Non-text part: multipart/alternative
> > Dear Christoph,
> >
> > On 24 August 2011 14:18, Christoph Helma <helma at in-silico.ch> wrote:
> >
> > > Dear Nina,
> > >
> > > I am not sure if Martin has replied already:
> > >
> >
> >
> >
> > >
> > > http://opentox.informatik.uni-freiburg.de/validation is the "official"
> > > validation service for all partners. For better load balancing we
> deploy
> > > additional services, e.g. webservices.in-silico.ch/validation for
> > > toxcreate.org,
> > > ot-test.in-silico.ch/validation for ot-test.in-silico.ch
> > > and most recently lazar-services.in-silico.ch/validation for
> > > lazar.in-silico.ch.
> > >
> >
> > Thank you.  As ToxPredict operates with arbitrary model URIs, what we the
> > best way to find out where the validation (and reporting ) resources for
> > particular model resides ? The question concerns not only models, built
> by
> > ToxCreate , but any models, registered in the ontology service (and thus
> > used by ToxPredict).
> >
> > Perhaps we should agree on some linking between model , validation and
> > reports, if these don't exist already?
>
> Yes. We are currently storing these associations within ToxCreate, but a
> common OpenTox mechanism would be very useful.  At present we can trace
> back from reports to validations to models (@Martin please correct me if
> I am wrong), but not the other way round.
>
> I can see the following options:
>
> - Register reports instead of models in the Ontology service (not very
>  intuitive)
>

We can store both, should be no problem.

Agree this should not be automatic, much like currently models are
registered manually, when necessary.


> - Add validation/report URIs to model metadata (at present our services
>  prohibit modifications of models after creation to avoid invalidation of
>  validation results)
>

That might be a bit tricky, as it will require write access to change model
metadata on a remote service.  In the general case, a validation (many
validations) can be run completely independent of models , with arbitrary
time delay.  And perhaps by using different validation services as well.


> - Store association between models, validations and reports at the
>  Ontology service
>


OK, to summarise, the most straightforward solution so far is to store the
validation and reports manually into the ontology service.
If validations and reports have RDF property pointing to a model URI,  one
can create SPARQL query to retrieve validations and reports , given a model
URI.

Will this work?

Best regards,
Nina


>
> Best regards,
> Christoph
>



More information about the Development mailing list