Re: updating SIDER LODD

Hi Peter,

Indeed the owl:sameAs relationship is not bidirectional. There are fewer
SIDER drugs map to DrugBank drugs because (1) the more regularly updated
DrugBank doesn't have an equivalent drug in SIDER (last update was in 2009,
so this is an issue with currency), and (2) even if a sameAs relation is
defined in the data graph, for some reason, the expected resulting graph
isn't generated.

For case #2 above, the query pulls a sider drug first and then assert the
sameAs relation with a drugbank drug. This results in the following graph
where the sider entries that are not correctly mapped to their equivalent
drugbank drug even though (if you actually click on the link and view the
sider drug's property) there exists a sameAs relationship in the data graph.

 125889 <http://www4.wiwiss.fu-berlin.de/sider/resource/drugs/125889>



 147912 <http://www4.wiwiss.fu-berlin.de/sider/resource/drugs/147912>

  222786 <http://www4.wiwiss.fu-berlin.de/sider/resource/drugs/222786>

How prominent is this problem with the LODD dataset?

Thanks!
Paul


PS: The sparql query is listed below...
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX sider: <http://www4.wiwiss.fu-berlin.de/sider/resource/sider/>
PREFIX drugbank: <
http://www4.wiwiss.fu-berlin.de/drugbank/resource/drugbank/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT  *
FROM <http://semanticweb.ics.uci.edu/LODD/SIDER#>
FROM <http://semanticweb.ics.uci.edu/LODD/DrugBank#>

WHERE {
?drug sider:sideEffect ?sideEffect .

# http://www4.wiwiss.fu-berlin.de/drugbank/resource/drugs/DB00230

OPTIONAL {
   ?ddDrug rdf:type drugbank:drugs;
      owl:sameAs ?drug .
  }
}

--
Paul Rigor
http://www.ics.uci.edu/~prigor



On Wed, Aug 3, 2011 at 4:07 PM, Peter Ansell <ansell.peter@gmail.com> wrote:

> Hi Paul,
>
> Given the difference in the bare numbers of drugs in each dataset, it
> is likely that the owl:sameAs relations are not all bidirectional.
> That may affect the semantics of both queries, as for one you are
> starting off in Sider and using its available owl:sameAs relations to
> get to DrugBank, and for the other you are starting off in Drugbank
> and using its owl:sameAs relations to get to Sider.
>
> Does your endpoint infer "X owl:sameAs Y" when it sees "Y owl:sameAs X"?
>
> You could do a simple test by comparing the results of:
>
> ?drug sider:sideEffect ?sideEffect .
> ?dbDrug rdf:type drugbank:drugs .
>    owl:sameAs ?drug .
>
> and
>
> ?drug sider:sideEffect ?sideEffect .
> ?dbDrug rdf:type drugbank:drugs .
>  ?drug owl:sameAs ?dbDrug .
>
> Cheers,
>
> Peter
>
> On 4 August 2011 06:36, Paul Rigor <paul.rigor@uci.edu> wrote:
> > Hi all,
> > As a follow up, here's some stats: There are over 4700 DrugBank drugs,
> 893
> > SIDER drugs.
> > More importantly, the equivalence mapping between SIDER and DrugBank drug
> > entities are not accurate (or not current). In fact, the mapping doesn't
> > seem to be symmetric. For example, if I query with a pattern where the
> > subject is a SIDER drug and the predicate a DrugBank drug, only 275
> overlap
> > between them. On the other hand, if the graph pattern uses a DrugBank
> drug
> > as the subject and the SIDER drug as a predicate, then resulting graph
> > captures a larger overlap with 833 drugs. Please let me know if I have
> > overlooked something! Thank you in advance.
> > The queries used are as follow:
> > (SIDER -> DrugBank)
> > PREFIX owl: <http://www.w3.org/2002/07/owl#>
> > PREFIX sider: <http://www4.wiwiss.fu-berlin.de/sider/resource/sider/>
> > PREFIX drugbank:
> > <http://www4.wiwiss.fu-berlin.de/drugbank/resource/drugbank/>
> > PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
> >
> > SELECT count distinct ?drug
> > FROM <http://semanticweb.ics.uci.edu/LODD/SIDER#>
> > FROM <http://semanticweb.ics.uci.edu/LODD/DrugBank#>
> >
> > WHERE {
> > ?drug sider:sideEffect ?sideEffect .
> > ?dbDrug rdf:type drugbank:drugs .
> >     owl:sameAs ?drug .
> > }
> > (DrugBank -> SIDER)
> > PREFIX owl: <http://www.w3.org/2002/07/owl#>
> > PREFIX sider: <http://www4.wiwiss.fu-berlin.de/sider/resource/sider/>
> > PREFIX drugbank:
> > <http://www4.wiwiss.fu-berlin.de/drugbank/resource/drugbank/>
> > PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
> > SELECT count distinct ?dbDrug
> > FROM <http://semanticweb.ics.uci.edu/LODD/SIDER#>
> > FROM <http://semanticweb.ics.uci.edu/LODD/DrugBank#>
> > WHERE {
> > ?dbDrug rdf:type drugbank:drugs .
> > ?drug sider:sideEffect ?sideEffect ;
> >      owl:sameAs ?dbDrug .
> > }
> >
> > --
> > Paul Rigor
> > http://www.ics.uci.edu/~prigor
> >
> >
> > On Wed, Aug 3, 2011 at 10:39 AM, Paul Rigor <paul.rigor@uci.edu> wrote:
> >>
> >> Hi gang,
> >> I was wondering whether there are plans to update the SIDER LODD and
> sync
> >> it with the latest version of the SIDER database? It's over two years
> old.
> >> Is there a published procedure for performing the RDFisation? I'm
> willing to
> >> help out as well.
> >> Cheers,
> >> Paul
> >>
> >> --
> >> Paul Rigor
> >> http://www.ics.uci.edu/~prigor
> >
> >
>