Dynamics, Structure and Evolution of Biological Networks Open Access
Downloadable ContentDownload PDF
We study the interplay between dynamics and structure of biological networks to gain understanding on their design principles and evolution pathways.We first studied issues in the structure inference of biological networks from binary time series data. A message-passing algorithm called belief propagation was brought to bear on the two inherent difficulties in Boolean network inference: low tolerance to noise perturbations and specific network selection from astronomically many alternatives. It provides accurate estimation about the likelihoods of individual interactions and the backbone network in roughly polynomial time. This new approach reconstructs large-scale biological systems under noise with high efficiency and accuracy.We then focus on the dynamical properties of complex networks and propose a new algorithm to detect the precursor of critical transition from high-dimensional time series data. Based on phase transition and percolation theory, the algorithm identifies a group of key variables with abnormal fluctuation and synchronization patterns that foretell a sudden change. The algorithm was applied for early detection of muscle regeneration and influenza symptoms.Lastly, efficient search algorithms are presented to analyze the exact genotype-phenotype maps and the precise topology of neutral network. The study distinguishes the discrepancy between individual and population, and therefore reconciles the paradoxical co-existence of robustness and evolvability in biological networks. The results also shed light on the different selection criteria of network evolution genotypically and phenotypically.