GQL: a reasonable complex SQL for genomic databases

Validating hypotheses and reasoning about objects is becoming commonplace in biotechnology research. The capability to reason strengthens comparative genomics research by providing a much-needed tool to pose intelligent queries in a more convenient and declarative fashion. To be able to reason using the Genomic Query Language (GQL), we propose the idea of parameterized views as an extension of SQL's "create view" construct with an optional "with parameter" clause. Parameterizing enables traditional SQL views to accept input values and to delay the computation of the view until invoked with a "call" statement. This extension empowers users with the capability of modifying the behavior of predefined procedures (views) by sending arguments and evaluating the procedure on demand. We demonstrate that the extension is soundly based, with a parallel in Datalog. We also show that the idea of relational unification proposed in this paper empowers SQL to reason and infer in exactly the same way as an object-oriented Datalog. Thereby, it eliminates the need for cumbersome integration of database engines with deductive reasoners, as was customary in many advanced genomic database applications in the past.