Many-core CPUs can deliver scalable performance to stochastic simulations of large-scale biochemical reaction networks

Stochastic simulation of large-scale biochemical reaction networks is becoming essential for Systems Biology. It enables the in-silico investigation of complex biological system dynamics under different conditions and intervention strategies, while also taking into account the inherent “biological noise” especially present in the low species count regime. It is however a great computational challenge since in practice we need to execute many repetitions of a complex simulation model to assess the average and extreme cases behavior of the dynamical system it represents. The problem's work scales quickly, with the number of repetitions required and the number of reactions in the bio-model. The worst case scenario s when there is a need to run thousands of repetitions of a complex model with thousands of reactions. We have developed a stochastic simulation software framework for many- and multi-core CPUs. It is evaluated using Intel's experimental many-cores Single-chip Cloud Computer (SCC) CPU and the latest generation consumer grade Core i7 multi-core Intel CPU, when running Gillespie's First Reaction Method exact stochastic simulation algorithm. It is shown that emerging many-core NoC processors can provide scalable performance achieving linear speedup as simulation work scales in both dimensions.