"JLIMS and its associated computational applications were designed to be robust enough to scale and meet this challenge, says Saul Kravitz, director of software engineering at the JTC. To achieve this scalability, JTC's data center employs an Oracle database running on Linux, and a collection of Linux servers.

To limit the manual scanning of bar codes and to impose process control, tight integration with fluid-handling robotic systems was required. Although commercial LIMS allowed integration via the reading and writing of control files for various instruments, they did not offer off-the-shelf integration directly with the instruments. Likewise, the center identified wireless personal digital assistants (PDAs) with integrated bar-code scanners as a key user interface, and the commercial offerings did not support these tools, according to Kravitz.

JLIMS is part of a larger set of applications responsible for processing, archiving, and delivering sequencing data, and providing operational and quality control reports. Using a commercial LIMS would have complicated integration with these computational and archival/retrieval components.

JLIMS addresses the need to interface to instruments, scalability, and the computation aspects of the application with a layered architecture comprised of PL/SQL stored procedures, a Java middle tier, and SOAP-based Web services.

JLIMS was developed in close collaboration between the software and lab teams. "The heavy investment in requirements and prototyping paid off with very few change requests following deployment," Kravitz says. The development process started with an examination of the needs of each lab. This involved interviewing lab personnel and observing lab procedures. The information gathered was then turned into a written assessment of each lab's requirements.

Next, prototypes of the graphical user interfaces (GUIs) were quickly developed using National Instruments' LabView. The GUIs were then demonstrated to the users and their feedback collected.

The developers then worked on the software. Once most of the functions were ready, an early version was released to a lab. The staff got to try the software for about a week, then provided additional feedback to help hone the application.

Once coding was complete, a validation process was carried out to ensure the software could function properly at the load levels required in a lab.

Back to LIMS: To Buy or Not to Buy?