Optimizing the Execution of Biological Models

Modelling biological systems allows us to understand how their components interact and give rise to complex behaviour. Initially, biology relied on mathematical models based on systems of differential equations whose solution describes the concentration of the molecules in time. Recently, spurred by the metaphor of “cells as computation†by Regev and Shapiro, the scientific community adapted concurrent languages to describe biological systems. This led to the creation of computational models which are executable and not simply solvable. Executable models offer some advantages over systems of differential equations, such as allowing the modeller to capture the causality relations among the events that constitute the dynamics of the model evolution. However, these new approaches introduce new issues; for example executing a model is more computationally intensive than solving a system of differential equations, especially if the model has to be executed several times because of statistical constraints. In this thesis we focus on reducing the execution time of biological models by applying static analysis techniques like control flow analysis and abstract interpretation.