Modularity-based approach identifies small sets of control nodes in synthetic and biological Boolean networks

This study investigates the problem of controlling the dynamics of biological systems to achieve desired outcomes (attractors). Specifically, we describe biological systems with Boolean models, which represent the system as a network, characterize system components (nodes) with binary states, and use discrete functions to describe the state changes of the nodes due to their interactions. We build upon the Feedback Vertex Set (FVS) theory, which guarantees that controlling the nodes in an FVS can drive the system toward a target attractor. However, FVS control can be computationally expensive in large networks. To overcome this, we propose two methods that exploit modularity within networks: (1) selecting the top 10% of influential nodes in each module based on structural metrics, and (2) identifying a subset of the FVS by prioritizing the highest-ranked nodes within each module. These approaches are evaluated in synthetic Random Boolean Networks (RBNs) and validated on real biological Boolean models. Our results show that modular control strategies are more efficient than global approaches, particularly in networks with clear modular structures, which is pertinent for controlling biological systems.