Indirect Targeting in Network Dynamics

Traffic jams, genetic diseases, financial systemic risk, ecosystem endangerment. These are some of the many examples of far-reaching imbalances in complex systems that can be caused by abnormal activity of one or few components­—a road, gene, financial institution, invasive species—which nevertheless are often resistant to direct control. Here, we develop a network solution to this broadly significant problem. We show that it is in general possible to manage, activate or even remove a node indirectly, through the exclusive manipulation of other nodes. Using two model systems, food-webs and gene regulatory networks, we show that indirect targeting is indeed achievable in networked systems. In particular, we show that we can eliminate an invasive species by acting on the other (e.g., native) species in food-web networks, and that the expression of a deleterious (or healthy) gene can be changed by acting on other genes in the regulatory network. Our approach exploits the structure of the phase space of the network, which allows the identification of feasible interventions that can lead to the desired activity of the target node while controlling for adverse effects. Our results indicate that indirect targeting can be accomplished with only temporary interventions and without relying on resources external to the network, permitting the design of focused, minimally intrusive control actions to modulate otherwise inaccessible critical nodes.