A cooperative framework for molecular biology database integration using image object selection

The theme and the concept of 'Molecular Biology Database Integration' and the problems associated with this concept initiated the idea for this Ph.D research. The available technologies facilitate to analyse the data independently and discretely but it fails to integrate the data resources for more meaningful information. This along with the integration issues created the scope for this Ph.D research. The research has reviewed the 'database interoperability' problems and it has suggested a framework for integrating the molecular biology databases. The framework has proposed to develop a cooperative environment to share information on the basis of common purpose for the molecular biology databases. The research has also reviewed other implementation and interoperability issues for laboratory based, dedicated and target specific database. The research has addressed the following issues: diversity of molecular biology databases schemas, schema constructs and schema implementation multi-database query using image object keying, database integration technologies using context graph, automated navigation among these databases. This thesis has introduced a new approach for database implementation. It has introduced an interoperable component database concept to initiate multidatabase query on gene mutation data. A number of data models have been proposed for gene mutation data which is the basis for integrating the target specific component database to be integrated with the federated information system. The proposed data models are: data models for genetic trait analysis, classification of gene mutation data, pathological lesion data and laboratory data. The main feature of this component database is non-overlapping attributes and it will follow non-redundant integration approach as explained in the thesis. This will be achieved by storing attributes which will not have the union or intersection of any attributes that exist in public domain molecular biology databases. Unlike data warehousing technique, this feature is quite unique and novel. The component database will be integrated with other biological data sources for sharing information in a cooperative environment. This involves developing new tools. The thesis explains the role of these new tools which are: meta data extractor, mapping linker, query generator and result interpreter. These tools are used for a transparent integration without creating any global schema of the participating databases. The thesis has also established the concept of image object keying for multidatabase query and it has proposed a relevant algorithm for matching protein spot in gel electrophoresis image. An object spot in gel electrophoresis image will initiate the query when it is selected by the user. It matches the selected spot with other similar spots in other resource databases. This image object keying method is an alternative to conventional multidatabase query which requires writing complex SQL scripts. This method also resolve the semantic conflicts that exist among molecular biology databases. The research has proposed a new framework based on the context of the web data for interactions with different biological data resources. A formal description of the resource context is described in the thesis. The implementation of the context into Resource Document Framework (RDF) will be able to increase the interoperability by providing the description of the resources and the navigation plan for accessing the web based databases. A higher level construct is developed (has, provide and access) to implement the context into RDF for web interactions. The interactions within the resources are achieved by utilising an integration domain to extract the required information with a single instance and without writing any query scripts. The integration domain allows to navigate and to execute the query plan within the resource databases. An extractor module collects elements from different target webs and unify them as a whole object in a single page. The proposed framework is tested to find specific information e.g., information on Alzheimer's disease, from public domain biology resources, such as, Protein Data Bank, Genome Data Bank, Online Mendalian Inheritance in Man and local database. Finally, the thesis proposes further propositions and plans for future work.