Noise dissipation in gene regulatory networks via second order statistics of networks of infinite server queues

Abstract

RNA and protein concentrations within cells constantly fluctuate. Some molecular species typically have very low copy numbers, so stochastic changes in their abundances can dramatically alter cellular concentration levels. Such noise can be harmful through constrained functionality or reduced efficiency. Gene regulatory networks have evolved to be robust in the face of noise. We obtain exact analytical expressions for noise dissipation in an idealised stochastic model of a gene regulatory network. We show that noise decays exponentially fast. The decay rate for RNA molecular counts is given by the integral of the tail of the cumulative distribution function of the degradation time. For proteins, it is given by the slowest rate-limiting step of RNA degradation or proteolytic breakdown. This is intuitive because memory of the chemical composition of the system is manifested through molecular persistence. The results are obtained by analysing a non-standard tandem of infinite server queues, in which the number of customers present in one queue modulates the arrival rate into the next.