Project Details

Aims

TAMBIS aims to provide transparent information retrieval and filtering from biological information services by building a homogenising layer on top of the different sources. This layer uses a mediator and many source wrappers to create the illusion of one all encompassing data source. Mediators

The mediator is a kind of information broker. It uses a conceptual model, an ontology, of molecular biology and bioinformatics, originally expressed in the Description Logic GRAIL and now in DAML+OIL, to

describe the universal model
help users form queries against this universal model expressed in the modelling language
mediate between the various sources to translate the mediators model to the sources’ models

The wrappers create the illusion of a common query language for each information resource. The current wrapper language is the Collection Programming Language (CPL) and BioKleisli and now uses the commercial version, DiscoveryHub, from GeneticXchange.

The mediator takes conceptual and source-independent GRAIL queries and rewrites them to concrete and source-dependent CPL queries, which are despatched to the sources.

The GRAIL encodings of the TAMBIS ontology are available:

The big model that covers Proteins, nucleic acids, their components, function, process, location, publishing and much more.
the baby model that covers only the protein subset of the big model. This was used for the "fully functioning version of the TAMBIS system.
the reconciled model that is the merged version of the big and baby TAMBIS ontologies.

The latest version of TAMBIS uses a translation of the original Baby TaO in GRAIL into the later DAML+OIL language. There is also an OWL (Web Ontology Language) version of this ontology.

The DAML+OIL version of Baby TaO.
The OWL version of Baby TaO.

On the way to building TAMBIS we have developed a number of other generally useful things including:

a A molecular biology and bioinformatics ontology
a questionnaire on bioinformaticians’ requirements
a benchmark query set

Biological Terminology Concept Server

The TAMBIS Ontology (TaO)is an ontology of molecular biological and bioinformatics concepts.

The Terminology Server provides a number of services for terminological or concept models such as what can I say about concept X? or is concept Y coherent? or what are the immediate parents and/or children of concept Z?

Together they form a Biological Terminology Concept Server used by TAMBIS.

The TaO is used to:

describe the metadata of the underlying data sources, representing an over-arching universal schema;
form queries expressed in the modelling language;
drive a GUI user interface for query formulation as users are not prepared to handle complex query languages;
mediate between the various sources to translate the mediators model to the sources’ models. Mediation exploits the TaO's lattice of concepts to assist in the identification and resolution of equivalences or near equivalences in information sources.

System architecture

The architecture of the TAMBIS system is shown in the figure below.
Architecture Diagram

The main components are:

A. An ontology of biological terms in DAML+OIL

B. A knowledge-driven query formulation interface

C. A Sources and Services Model linking the biological ontology with the source schemas

D. A query transformation rewriting process

E. A wrapper service dealing with external sources (DiscoveryHub)

The User Interface

Queries are constructed using a graphical query interface which allows users to piece together queries and insulates them from the syntax of the underlying description logic model. An applet demonstrating the TAMBIS system is no longer accessible.

A series of videos are available:

The user manual for the TAMBIS applet gives full details of how to use TAMBIS.