Overview of Biological Network Inference and Modeling of Dynamics

Abstract

In the era of high-throughput experiments, inferring and modelling the dynamics of biological systems are complex tasks. The complexity derives from the large sizes, the presence of competing interactions, stiffness, and non-linearity in the systems under investigation. Moreover, the dynamics in these systems are typically hybrid — that is, stochastic and deterministic and time irreversible — raising many technical and conceptual challenges: is it possible, at least in principle, to infer the topology and the properties of a biological network from observations of the dynamics? How much and what information do we need to obtain from an experiment to accurately infer a network model? Would it be possible to accurately describe the dynamics of the stochastic interactions without a full stochastic simulation, which is extremely computationally expensive for systems of huge size? How is it possible to infer from partial and noisy observations the trajectory of the system for complete observations? All these questions are born of the realistic possibility that the inference and modeling of omic-size, complex interaction networks is an underdetermined problem. In this introductory chapter, we present the challenges underdetermination modeling has to face in systems biology and its usefulness in the era of high-throughput experiments and big data collection.