Correspondence

Agile methods in biomedical software development: a multi-site experience report

David W Kane^{* 1} , Moses M Hohman^{* 2} , Ethan G Cerami³ , Michael W McCormick⁴ , Karl F Kuhlmman⁵ and Jeff A Byrd⁶

¹SRA International, 4300 Fair Lakes Court, Fairfax, VA 22033, USA

²Center for Functional Genomics, Northwestern University, 2205 Tech Drive #2-160, Evanston, Illinois 60208, USA

³Memorial Sloan-Kettering Cancer Center, Computational Biology Center, 1275 York Avenue, Box #460, New York, NY 10021, USA

⁴Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, Seattle, WA 98109, USA

⁵Applied Biosystems, 850 Lincoln Centre Drive, Foster City, CA 94404, USA

⁶Vanderbilt Medical Center, 2209 Garland Avenue, 416 Eskind Biomedical Library, Nashville, TN 37232-8340, USA

author email corresponding author email* Contributed equally

BMC Bioinformatics 2006, 7:273doi:10.1186/1471-2105-7-273

Published:	30 May 2006

Abstract

Background

Agile is an iterative approach to software development that relies on strong collaboration and automation to keep pace with dynamic environments. We have successfully used agile development approaches to create and maintain biomedical software, including software for bioinformatics. This paper reports on a qualitative study of our experiences using these methods.

Results

We have found that agile methods are well suited to the exploratory and iterative nature of scientific inquiry. They provide a robust framework for reproducing scientific results and for developing clinical support systems. The agile development approach also provides a model for collaboration between software engineers and researchers. We present our experience using agile methodologies in projects at six different biomedical software development organizations. The organizations include academic, commercial and government development teams, and included both bioinformatics and clinical support applications. We found that agile practices were a match for the needs of our biomedical projects and contributed to the success of our organizations.

Conclusion

We found that the agile development approach was a good fit for our organizations, and that these practices should be applicable and valuable to other biomedical software development efforts. Although we found differences in how agile methods were used, we were also able to identify a set of core practices that were common to all of the groups, and that could be a focus for others seeking to adopt these methods.