MDB: a database system utilizing automatic construction of modules and STAR-derived universal language

The value of information greatly increases if stored in databases. The objective was to construct a multi-purpose database system primarily designed to store and provide access to three-dimensional structures of biological molecules including theoretical models.

A dictionary defining data format and structure for three-dimensional models of biological molecules (MDB dictionary) was developed. The dictionary was written using universal, standardized data description language. This language can be applied to describe data with no restrictions on their origin or type, including metadata. Thus both the data definitions (format) and database descriptions are created using the uniform language and processed with universal software. A database and data design technique that allowed use of dictionaries to automatically construct relational databases was developed. This technique was employed to construct the MDB database system. Data design developed and applied in the MDB project makes it possible to carry out data curation utilizing the database engine to identify errors. It also allows storage and query of data at different levels of consistency with the standard format specifications, i.e. both the correctly formatted data, and data that requires further curation.

The MDB dictionary is available at http://www.gwer.ch/proteinstructure/mdb and as part of the PDB resources at http://pdb.rutgers.edu/mmcif/.