Stochastic Multiple-Valued Gene Networks.

Peican Zhu,Jie Han

DOI: https://doi.org/10.1109/tbcas.2013.2291398

2014-01-01

IEEE Transactions on Biomedical Circuits and Systems

Abstract:Among various approaches to modeling gene regulatory networks (GRNs), Boolean networks (BNs) and its probabilistic extension, probabilistic Boolean networks (PBNs), have been studied to gain insights into the dynamics of GRNs. To further exploit the simplicity of logical models, a multiple-valued network employs gene states that are not limited to binary values, thus providing a finer granularity in the modeling of GRNs. In this paper, stochastic multiple-valued networks (SMNs) are proposed for modeling the effects of noise and gene perturbation in a GRN. An SMN enables an accurate and efficient simulation of a probabilistic multiple-valued network (as an extension of a PBN). In a k-level SMN of n genes, it requires a complexity of O(nLk(n)) to compute the state transition matrix, where L is a factor related to the minimum sequence length in the SMN for achieving a desired accuracy. The use of randomly permuted stochastic sequences further increases computational efficiency and allows for a tunable tradeoff between accuracy and efficiency. The analysis of a p53-Mdm2 network and a WNT5A network shows that the proposed SMN approach is efficient in evaluating the network dynamics and steady state distribution of gene networks under random gene perturbation.