UC Merced

It is challenging to build meaningful models of biological systems that are calibrated to experimental data. For microbial colonies, chromogenic assays are key tools for distinguishing phenotypic variants in growing colonies over time and provide additional useful information for colony quantification. In microbial colonies, the presence of multiple phenotypes indicate a functional change that is dependent on the colony species. One example in Saccharomyces cerevisiae is that the presence of multiple phenotypes indicates a loss of infectious agents within the cells. Another example in Candida albicans is that the presence of multiple phenotypes tells us about cell mating strategies. At present, we lack both a model that explains the formation of these sectored phenotypes and a method for validating such a model from experiments. In addition, we lack a framework which couple both approaches to make meaningful insights related to the driving of multiple phenotypes in microbial colonies. In this dissertation, I seek to address this gap with a data-driven approach that integrates experimental data of sectored yeast colonies with the construction of an agent-based model of growing budding yeast colonies. I first discuss my previously published work developing an agent based model of budding yeast and apply mathematical techniques to analyze colony structure. Next, I explain my ongoing work to use image analysis techniques to create a high-throughput pipeline allowing us to extract information about the size and structure of colonies found in image data. Finally, I present ongoing work with using this framework for larger and more diverse datasets and discuss limitations on their use in automated colony quantification. Together these projects create a framework allowing us to adapt our agent based modeling framework to the conditions observed in experiments so that the model both captures sectoring behavior and allows us to predict the mechanisms driving sectoring behavior.