[Back to online resume.]

John K. Hinsdale

Work Experience: Bristol-Myers Squibb Company

SUMMARY OF SOFTWARE PROJECTS

As contractor to Bristol-Myers Squibb I designed and worked extensively with two systems: Both systems are considered mission-critical and are still in use today, with several hundred users each.

Project: DNA Sequencing Automation

A Web-based database for tracking DNA sequencing research, automating all stages of the sequencing process: logging and submission of DNA samples, workflow/queue management for lab equipment operators, collection and routing of output data from DNA sequencing robots, post-processing ("crunching") of results, and E-mail notification back to the end users, including error checking and data quality statistics.

The highly-automated nature of the system allows the research group to sequence thousands of samples a day at each of three locations.

DNA SEQUENCING AUTOMATION: Project Summary
Timeline:1997 - 1998, 2000 - present
Technologies: perl, Oracle, Informix, Apache, Sun/Solaris, SGI/Irix, Linux, CVS
Contribution: Requirements gathering, complete data modelling, software design, implementation, test, documentation, deployment and enhancement.
Innovations: The portability of perl and SQL proved very useful in this project, because by necessity it was developed in one location and deployed in another. The development environment used Sun/Solaris (and later, Linux) with an Informix database and Apache Web server, while the system itself was deployed on SGI/Irix with an Oracle database and Netscape Web server. The same software ran in each environment without modification.
Lessons learned: We looked carefully at what activities users were doing, and listened to exactly what they wanted the system to automate, then designed the database from scrach around that. We thus avoided a very common failing of off-the-shelf LIMS systems: they try to solve a too-general problem, without ever quite meshing with the activities they attempt to automate. Here, the tight alignment between the application and the actual workflow was key to the success of the system.

Project: Genomic Curation

A Division-wide repository for experiment results and annotation of sections of the DNA sequence of the Human Genome. The system is geared toward searching and archival, including revision tracking of every update, built-in audit trails, and document integration capabilities between in-house proprietary data and public data. System allows upload file attachments, auto-linking to public external sites, user usage reports, and many other features.

The system has evolved into the research division's principal repository for experimental intellecutal property (IP) and is now used by about three hundred researchers.

GENOMIC CURATION: Project Summary
Timeline:2000 - present
Technologies: perl, Oracle, LDAP, Apache, SGI/Irix, Linux, CVS
Contribution: Requirements gathering, complete data modelling, software design, implementation, test, documentation, deployment and enhancement.
Innovations: We could not predict what kinds of data would eventually be stored, so we made an extensible data model that allowed users to add new fields "on the fly." We made an API into the database so that other systems could feed into it, increasing its value as an integration tool.
Lessons learned: A legacy system was available, and use of a model very similar to that helped to get users to adjust to the new system more smoothly. Since the database was not just a repository but a permanent archive, we focussed on having a robust, but simple implementation from the outset to increase user confidence.

[Back to online resume.]