Inferring signaling pathway topologies from multiple perturbation measurements of specific biochemical species.

The specification of biological decisions by signaling pathways is encoded by the interplay between activation dynamics and network topologies. Although we can describe complex networks, we cannot easily determine which topology the cell actually uses to transduce a specific signal. Experimental testing of all plausible topologies is infeasible because of the combinatorially large number of experiments required to explore the complete hypothesis space. Here, we demonstrate that Bayesian inference-based modeling provides an approach to explore and constrain this hypothesis space, permitting the rational ranking of pathway models. Our approach can use measurements of a limited number of biochemical species when combined with multiple perturbations. As proof of concept, we examined the activation of the extracellular signal-regulated kinase (ERK) pathway by epidermal growth factor. The predicted and experimentally validated model shows that both Raf-1 and, unexpectedly, B-Raf are needed to fully activate ERK in two different cell lines. Thus, our formal methodology rationally infers evidentially supported pathway topologies even when a limited number of biochemical and kinetic measurements are available.