Noise and reliability in Boolean networks

Gene regulation is an inherently stochastic process, which often needs to function with some degree of predictability, despite the presence of ubiquitous fluctuations. In this talk, we use Boolean networks as a model for gene regulation, and establish what general properties are necessary -- both functional and topological -- so that their dynamics are robust against noise. We consider two general types of noise: 1. Fluctuations in the update sequence of genes; and 2. Fluctuations in their expression level. We show that it is possible for a network to be entirely resilient against the first type of noise, and describe the functional and topological characteristics which are necessary to achieve this. We then consider networks with functional redundancy, which are conditioned for robustness against the second type of noise. Unlike the previous type of noise, we show it is not possible for a network to be entirely reliable against fluctuations in the expression level. We obtain upper bounds on the attainable reliability due to redundancy, if the number of inputs per node is finite and fixed. We also show that there is always a maximum amount of noise, above which redundant networks are no longer resilient, and the system undergoes a transition from non-ergodic to ergodic behavior.

Return