Inference of Granger-causal relations in molecular systems — a case study of the functional hierarchy among actin regulators in lamellipodia
Many cell regulatory systems implicate significant level of nonlinearity and redundancy among its components. The actin regulatory network driving the formation of a lamellipodium is prototypical of such system, containing tens of actin nucleating and modulating molecules with strong functional overlap. Due to instantaneous compensation, the standard paradigm of perturbing individual components followed by phenotyping provides limited information on the roles the targeted component plays in the unperturbed system. Accordingly, despite the very rich data on lamellipoidial actin assembly, we have an incomplete understanding of the actual contributions the individual modulators make towards the lamellipodial dynamics. Here, we present a case study of the implementation of Granger-causal inference of the functional cause-effect relations among actin regulators, using the constitutive image fluctuations reporting regulator recruitment/activation as the input. Our analytical pipeline defines specific active regions of actin regulators within the lamellipodia and lamella and establishes actin-dependent and actin-independent causal relations of actin regulators with F-actin and edge motion. We demonstrate the specificity and sensitivity of the pipeline and suggest the existence of two discrete yet independently operating F-actin networks that drive edge motion.