(362d) Maximum Entropy Approach for Parameter Estimation in Signaling Networks

Predictive models of signaling networks are key to our understanding of cellular function and in designing rational interventions in disease. However, using network models to predict signaling network behavior is challenging due to inherent cell-to-cell variability of network parameters, such as reaction rates and species concentrations that govern network dynamics. In this work, we present an inference framework based on the principle of maximum entropy (ME) to estimate the joint probability distribution over network parameters that is consistent with experimentally observed cell-to-cell variability in concentrations of network species. We apply the framework to study the signaling cascade activated by the epidermal growth factor (EGF) resulting in phosphorylation of protein kinase B (Akt); a central signaling hub in mammalian cells. Notably, the inferred parameter distribution allows us to computationally predict the distributions of phosphorylated Akt (pAkt) levels at early and late times after EGF stimulation as well as the distribution of cell surface EGF receptors (sEGFRs) after prolonged stimulation with EGF. We discuss how the developed framework can be generalized and applied to other types of data as well as problems beyond signaling networks.