A Universal Minimum Description Length-Based Algorithm for Inferring the Structure of Genetic Networks

The Boolean network paradigm is a simple and effective way to interpret genomic systems, but discovering the structure of these networks is a difficult task. In this paper, we model genetic time series data as multivariate Boolean regression and employ the minimum description length principle to find significant relationships among the genes. The description length is based upon a universal normalized maximum likelihood model, and we use an analogue of Kolmogorov's structure function to reduce computation time. The performance of the proposed method is demonstrated on random synthetic networks.