[工程科技]Executive Summary Findings.pdfVIP

Practical Software Engineering Techniques for Regulatory Models in Biology James Watson, Jim Hanan and Janet Wiles ARC Centre for Complex Systems The University of Queensland Email: jwatson@.au Executive Summary Findings This report has focused on verification and validation techniques that are suited to GRN research methods and are of immediate, practical benefit to the computational modelers of the ACCS community. Through an online survey and focused interviews, this project identified the following key char- acteristics of GRN modeling: 1. Small team size: Simulations are implemented and used by a very small number of people, often a single developer is the single user. 2. Transient systems: Software models are developed to help under- stand a particular research question, and are often redesigned or even discarded as the question / understanding changes. 3. Rapidly evolving specifications: Simulations are developed and used in an iterative situation of rapid specification change, with speci- fications often changing as quickly as the software is run. 4. Non-linear: The precise behaviour of the complete simulation is un- known before runtime, with emergent interactions of system compo- nents providing the observations of interest. Recommendations The rapidly evolving requirements coupled with small team sizes (points 1-3 above) conspire to make many popular software engineering techniques impractical. The following recommendations distil the most useful techniques for such modeling: A Maintenance of static components: Although the specifications of such software systems undergo rapid change, many components remain i the same, a