A Database-Centric Virtual Chemistry System

We describe an Oracle database application for general use within virtual chemistry. The application functions
as a central hub and repository for chemical data with interfaces to external calculators. It deals with the
general problems of merging data from disparate sources and with scheduling of computational tasks for
parallel or sequential execution in a mixed environment. The central database is used for the storage of
input, intermediary, and final data as well as for job control. A calculation job is split into distinct tasks, or
units of work, which are put in a queue. Tasks are dequeued and handled by specialized calculators. These
calculators are in-house or commercial programs for which adaptor modules for connection to the database
must be written. Tasks are handled in a transactional fashion, so that uncompleted or failed tasks are left in
the queue. This makes the system stable to many types of disturbances. Sorting, filtering, and merging
operations are handled by the database itself. Usage is very general, but some specific examples are (1) as
a back end for a chemical property calculator Web page, (2) in an automated quantitative structure−activity
relationship system, and (3) in virtual screens.