Title: Integrating Protein Data Resources through Semantic Web Services

When: November 7, 2006  10:00am – 12:00pm

Where: Room 522 IS Bldg.

Who: Xiong Liu

Committee:
Dr. Hassan Karimi (advisor), School of Information Sciences, University of Pittsburgh
Dr. Michael Lewis, School of Information Sciences, University of Pittsburgh
Dr. Vladimir Zadorozhny, School of Information Sciences, University of Pittsburgh
Dr. Ivet Bahar, Department of Computational Biology, School of Medicine, University of Pittsburgh
Dr. John Vries, Department of Computational Biology, School of Medicine, University of Pittsburgh

Abstract:Understanding the function of every protein is one major objective of bioinformatics. Currently, a large amount of information (e.g., sequence, structure and dynamics) is being produced by experiments and predictions that are associated with protein function. Integrating these diverse data about protein sequence, structure, dynamics and other protein features allows further exploration and establishment of the relationships between protein sequence, structure, dynamics and function, and thereby controlling the function of target proteins. However, information integration in protein data resources faces challenges at technology level for interfacing heterogeneous data formats and standards and at application level for semantic interpretation of dissimilar data and queries.

In this research, a semantic web services infrastructure, called Web Services for Protein data resources (WSP), for flexible and user-oriented integration of protein data resources, is proposed. This infrastructure includes a method for modeling protein web services, a service publication algorithm, an efficient service discovery (matching) algorithm, and an optimal service chaining algorithm. Rather than relying on syntactic matching, the matching algorithm discovers services based on their similarity to the requested service. Therefore, users can locate services that semantically match their data requirements even if they are syntactically distinctive. Furthermore, WSP supports a workflow-based approach for service integration. The chaining algorithm is used to select and chain services, based on the criteria of service accuracy and data interoperability. The algorithm generates a web services workflow which automatically integrates the results from individual services.

A number of experiments are conducted to evaluate the performance of the matching algorithm. The results reveal that the algorithm can discover services with reasonable performance. Also, a composite service, which integrates protein dynamics and conservation, is experimented using the WSP infrastructure.