Jamie Twycross: Principled Selection of Stochastic Simulation Algorithms

Stochastic simulation algorithms (SSAs) allow biologists to perform accurate simulations of chemical systems at low species concentrations. As the complexity of models increases, this computationally expensive algorithm can become intractable. Numerous improvements to SSAs have been introduced but they each tend to only apply to a certain class of model. In my talk I will give an overview of a paper in preparation. In the paper we investigate whether it is possible to determine which algorithms are suited to a particular class of model, and if this can be ascertained a priori to simulation. To answer this question we have developed an SSA performance benchmarking suite. The suite provides reference implementations of the main classes of SSAs. We also provide a set of models, classified using a number of metrics. We compare the computational efficiency of each implementation on these test cases to determine which SSAs perform best on different classes of biological reaction networks.