Parallel stochastic simulation using graphics processing units for the Systems Biology Toolbox for MATLAB Software usage guide

This document gives a brief guideline on how to use the parallel stochastic simulation of chemical reaction systems on graphics processing units (GPUs) plugin for MATLAB (STOCHSIMGPU plugin). The plugin is tightly integrated into the System Biology Toolbox 2 for MATLAB (SBTOOLBOX2) [7]. The STOCHSIMGPU plugin is a direct replacement for the stochastic simulation algorithm provided by the SBTOOLBOX2 but provides additional features. These are the implementation of three exact stochastic simulation algorithms, Gillespie’s stochastic simulation algorithm (SSA) [3], Li and Petzold’s logarithmic direct method (LDM) [4] and the next reaction method (NRM) by Gibson and Bruck [2] and the histogram computation across all realisations of a stationary process, the steady state or at any user defined point in time. These additional features are available through additional optional parameters given to the SBstochsim script. The STOCHSIMGPU plugin is a direct substitute and no change to the user’s code is required. The usage of the stochastic simulation within the SBTOOLBOX2 is given in documentation at http://www.sbtoolbox2.org. The basic background in parallel computing on a GPU and the implemented exact stochastic simulation algorithms is given in the supplemental material.