Bioinformatics reactivity features through the Semantic Web

         Tiago Franco and José Alferes and Ludwig Krippahl and Ricardo Amador
           CENTRIA - Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa

Abstract
In order to propagate changes in, and between information nodes in the Web, a high-level
architecture has been proposed for a reactive framework aligned with the goals and technolo-
gies of the Semantic Web – r3 (see http://rewerse.net/I5/r3/). As part of the validation of this
framework, we have used Bioinformatics as a case study. This approach required the design
and implementation of a service ontology for the Bioinformatics domain (B-Domain). The
ontology describes both the static and dynamic concepts of the domain, enabling reasoning
and rule composition, without disregarding language heterogeneity in the Semantic Web.




        Fig. 1. Validation Scenario: inferring the presence of Dimers in a new PDB structure

    The concept was validated with 4 scenarios similar to the one presented in Fig. 1. This
particular scenario shows a user being notified whenever a new dimer is published in PDB.
To accomplish this task the scenario depicts the use of the B-Domain and SPARQL Brokers
and of a Personalized Mirror. The first component integrates information sources relevant to
bioinformatics exposing its features through B-Domain’s ontology. The second leverages the
features of the SPARQL query language to the framework’s Enterprise Service Bus. The Event
Condition Action (ECA) Engine is an r3 rule engine that allows the registration and execu-
tion of Event-Condition-Action rules. Finally, the Personalized Mirror is a simple component
that allows users to store RDF/XML data. The external interfaces of each component are de-
scribed using the r3 framework’s ontology. The execution of the validation scenarios showed
that Bioinformatics would strongly benefit from the features provided by B-Domain. Its char-
acteristics allow the description of the static and dynamic semantic features of the domain, en-
abling reasoning on top of those concepts. It also offers a way to create and deploy declarative
rules (ECA), which can be chained to create complex meaningful rules. The running version
of B-domain can be found at http://rewerse.net/I5/r3/TST/install/dev/other/b-domain/.