Developing Agent-Based Models for Bacterial Chemotaxis with Sonification-Aided Data Collection

Bacterial motility and navigation are fundamental to a number of biological processes, from infection to bioremediation. Critical to bacterial navigation is chemotaxis, the ability of bacteria to respond to chemical cues around them by swimming towards or away from their sources. Chemotaxis in populations of bacteria is often modeled using diffusion-convection continuum equations, but capturing stochasticity, heterogeneity, and complex environmental factors using these mathematical models can be non-trivial. In this work, I explore how we can enhance bacterial chemotaxis research through sonification, and by using agent-based models to simulate bacterial systems with complex rules.
Sonification is the act of communicating or expressing data in the form of sound, just as visualization of data uses visual formats such as graphs and images. Sonification is used by scientists to identify patterns in noisy data and to assist with process monitoring tasks. In this dissertation, I apply sonification to microscope video and image output in order to provide a secondary source of information to communicate visual features observed through the microscope that are relevant to chemotaxis research as a form of process monitoring. This approach would allow researchers to set up dynamic chemotaxis assays and monitor their progress via sound while attending to other tasks. I describe the implementation of this sonification platform and show how features of the sound are related to features of interest in the microscopy videos. I then evaluated the platform by administering a user survey and demonstrated that users could effectively relate features of the sound to the microscope data to correctly identify a chemotactic response and a moving wave of chemotactic bacteria in the videos. This sonification-supported process monitoring could make the collection of videos of bacteria chemotaxis more efficient.
Agent-based models are a type of simulation that represents individual agents with specific rules to obtain emergent behaviors. To model systems with complex rules, I developed an agent-based model of bacterial chemotaxis and validated it with experimental data. I then developed a platform for determining chemotaxis parameters for the agent-based model by analyzing trajectories of chemotactic bacteria. I validated this platform with simulated tracks, and then demonstrated its use on experimental data of E. coli and α-methyl aspartate to assess chemotaxis parameters for this bacteria-chemoeffector pairing. Then, I used the agent-based model to explore a system that used a microfluidic device to expose bacteria under flow to a source of chemoattractant. Experimental observations indicated that the chemotactic response was reduced as the flow rate increased, but a continuum model of the system did not provide insight to the mechanisms underlying the observation. I used the agent-based model to test several hypotheses to explain the observed decrease in chemotactic response and proposed that shear flow within the system hampering bacterial motility would be a potential explanation for decreased chemotaxis. In future work, the agent-based models could be adapted to simulate more complex scenarios in order to probe the role of chemotaxis in groundwater remediation and similar contexts.