TAMBIS

1. TAMBIS Transparent Access to Multiple Biological Information Sources

2. Why Tambis? • TAMBIS aims to provide transparent information retrieval and filtering from biological information sources. • This will be through the use of a homogenising layer on top of the sources. • The layer uses a mediator and wrappers to create the illusion of a single data source.

3. Wrapper Wrapper Wrapper Mediators • The mediator is an information broker. It uses a conceptual knowledge base of biology to: • Describe a universal model Mediator Mediator • Help users form queries • Translate the mediator’s model to the sources’ model

4. Wrappers • Wrappers create the illusion of a common query language for each information resource. Mediator • This insulates the mediator from differences in source access methods Wrapper Wrapper Wrapper Wrapper Wrapper Wrapper • The current wrapper language is CPL

5. Terminology Server Biological Terminology Server Query Formulation Dialogues Services KB Query Transformation Wrapper Service Terminology Server Biology Concept Model Linguistic Model Architecture • The Terminology Server provides services for reasoning about concept models, answering questions like: What can I say about Proteins? What are the parents of concept X? • It communicates with other modules through a well-defined interface

6. Query Formulation Dialogues Biological Terminology Server Query Formulation Dialogues Services KB Query Transformation Wrapper Service Architecture • The user interacts with Query Formulation Dialogues, expressing queries in terms of the biological model. • The dialogues are driven by the content of the model, guiding the user towards sensible queries. • The query is then passed to the transformation process, which may require further user input to refine and instantiate the query.

7. Services KB Biological Terminology Server Query Formulation Dialogues Services KB Query Transformation Wrapper Service Services KB Concepts  SSM mapping BISSMap Source Combination Model SCM Source and Services Model SSM Architecture • The Services Knowledge Base links the biological ontology with the sources and their schemas. • This information is used by the transformation process to determine which source should be used.

8. Query Transformation Biological Terminology Server Query Formulation Dialogues Services KB Query Transformation Wrapper Service Architecture • Query Transformation takes the conceptual source-independent queries and rewrites to produce executable query plans. • To do this it requires knowledge about the biological sources and the services they offer. • Information about particular user preferences - say favourite databases or analysis methods - may also be incorporated by the query planner. • The query plans are then passed to the wrappers.

9. Wrapper Service Biological Terminology Server Query Formulation Dialogues Services KB Query Transformation Wrapper Service Wrapper Service Query Execution Coordinator Wrapper Client Wrapper Client Wrapper Client Architecture • The Wrapper Service coordinates the execution of the query and sends each component to the appropriate source. • Results are collected and returned to the user.

10. Modelling Biology with DLs • The Biological Concept model is built using a Description Logic or DL. • Primitive concepts are atomic terms, e.g. Protein or Motif. • Roles denote binary relationships between concepts, e.g. hasOrganismSource, isComponentOf. • Term constructors associate concepts and roles to define composite concepts, e.g. Motif which isComponent of Protein. • Concepts are both definitions that form the model and queries on the model - the same language is used.

11. SequenceComponent Motif which <isComponentOf (Protein which hasOrganismSource PoeciliaReticulata) hasFunction Hydrolase> SequenceComponent which hasFunction Hydrolase SequenceComponent which isComponentOf Protein Motif which hasFunction Hydrolase Motif which isComponentOf Protein Motif Modelling Biology with DLs • Primitive concepts are placed by the modeller into a subsumption (or kind-of) hierarchy. • Composite concepts are automically classified in the hierarchy based on the description of the concept.

12. Modelling Biology with DLs • The combination of concepts with roles is tightly controlled. We use these controls together with the classification to check the coherency of a concept. • Two concepts are permitted to be related via some role through the use of sanctions. Composite concepts can’t be formed without sanctioned permission. • Motif isComponentOf Protein • NucleicAcidComponent isComponentOf Protein • Sanctions ensure that only semantically valid compositions are formed; a large number of compositions can be inferred from a sparse model. • They also allow us to answer questions like “what can I say about this concept?”

13. Graphical presentation of Query Hierarchical view of parent concepts TAMBIS in action Query Interface

14. Accept query TAMBIS in action Query expression Motif which is ComponentOf (Protein which hasOrganismSource PoeciliaReticulata) Rewrite to CPL {motif1 | \protein1 <- get-sp-entry-by-os(“POECILIA+RETICULATA”), \motif1 <- do-prosite-scan-by-entry-rec(protein1)} Evaluate query Query Interface

15. TAMBIS is a collaboration between the departments of Computer Science &Biological Science at the University of Manchester, funded by EPSRC and Zeneca Pharmaceuticals To find out more contact: tambis@cs.man.ac.uk • Andy Brass • Biochemistry and Molecular Biology,University of Manchester,Oxford Road,Manchester M13 9PL, UKPhone: +44 161 275 5064/5096Fax: +44 161 275 5082 • E-mail: abrass@manchester.ac.uk • Carole Goble • Computer Science,University of Manchester,Oxford Road,Manchester M13 9PL, UKPhone: +44 161 275 6195Fax: +44 161 275 6932 • E-mail: carole@cs.man.ac.uk • Norman Paton • Computer Science,University of Manchester,Oxford Road,Manchester M13 9PL, UKPhone: +44 161 275 6910Fax: +44 161 275 6932 • E-mail: norm@cs.man.ac.uk