An Agent-Based Model of Spatially Structured Bacterial Populations

Although more computationally intensive, agent-based models have many advantages over alternative simulation approaches. In particular, stochastic processes and the resulting emergent behaviors can be modeled more realistically. Here, we discuss the Stochastic Agent-Based Network-Fixation Computed Topology with Undirected Migration (SANCTUM) model. In it, agents are given qualitative or quantitative "phenotypes" and placed in one of a set of patches connected by migration paths. Each patch has a fixed number of available spaces, as well as specific properties, like antibiotic concentration. Agents undergo growth, death, and migration steps stochastically based on their phenotype and patch characteristics. Sigmoidal growth and resource limitations previously modeled using the Lotka-Volterra equations instead emerge naturally based on the number of unoccupied spaces within a patch. Using SANCTUM, we can assess the impact of network topology on the chance that a rare mutation - for example, one that confers increased antibiotic resistance - will take over a population. In this way, the evolutionary pressures that act on heterogeneous populations can be explored more systematically. Future work can extend the model to test the effectiveness of "drug sanctuaries," in which the use of certain drugs is restricted for certain patches and/or times to suppress the emergence of resistance. Similar work can be done for proposed ‘adaptive therapy’ regimes to treat cancer which seek to maximize the effectiveness of chemotherapy or radiation by carefully controlling selection pressures. Thus, SANCTUM can serve as a platform for in silico testing of novel treatment approaches based on understanding the underlying population and evolutionary dynamics.