Unbiased τ-leap methods for stochastic simulation of biochemical systems

Stochastic simulation of biological systems has received much attention recently. A very promising stochastic simulation method is the tau-leap method, which can significantly accelerate simulation with controllable accuracy. However, all current -leap methods produce biased results, which can cause large simulation errors. In this paper, we analyze the expected number of reactions occurring during each leap. Relying on the analytical results, we develop an unbiased Poisson tau-leap method and an unbiased binomial tau-leap method. Simulations demonstrate that our new unbiased tau-leap method can significantly improve simulation accuracy without sacrificing simulation speed.