Scientific data repositories are often isolated and characterized by various degrees of heterogeneity: they usually represent different views (schemas) of the scientific domain, where the schemas are specified using different languages (data models), and are implemented using different data management systems, such as file management systems or database management systems (DBMSs). Interacting with multiple data repositories requires the ability of understanding and correlating schemas specified using different data models.
Data models currently used for describing molecular biology data repositories include (but are not limited to): the relational model (underlying all relational DBMSs), various versions of the Entity-Relationship Model (ERM) [3] and Extended Entity-Relationship Model (EERM) [13], the data model of Genera [11], the data model of ACEDB (underlying the ACEDB data management system) [5], and the Object-Protocol Model (OPM) [4].
For example,
the relational model is used for specifying (directly or via tools)
the databases implemented with relational DBMSs, such as
the Genome Sequence Data Base (GSDB) and Genome Data Base (GDB),
EER is used for specifying the object schema Flybase database
at Harvard University and several databases at the Chromosome 22
Center at University of Pennsylvania;
Genera is used for specifying the object schema of
the Enzyme database
and the XLocus database of cross-species gene-based linkages;
ACEDB is used for developing numerous databases, including
the Integrated Genomic Database (IGD);
OPM is used for specifying the object schema of databases such as the
Citations database,
of the new GDB multidatabase system,
and the Protein Data Bank (PDB).
The data models of Genera, ACEDB, EERM, and OPM share similar constructs. We briefly examine these data models using a reference data model and discuss the problem of converting schemas between these data models.