Grant: Discovery Grants Program - Individual

Microbial communities play fundamental roles in human and animal health, agriculture, environmental remediation, and biomanufacturing. Consequently, targeted manipulation of these communities holds promise in a wide range of areas. The long-term goal of my research program is to support such efforts by providing novel computational techniques for the development and analysis of predictive models of the population dynamics of microbial communities. Such communities exhibit a staggering array of behaviours and compositions. Rather than tackle this diversity directly, we take a bottom-up approach, starting with accurate predictive characterization of lab-scale mixed microbial populations. Specifically, the proposed work will focus on bacterial growth, gene transfer, mutualism, and antagonistic interactions. The resulting models will be validated against experimental observations of idealized communities grown in simple laboratory environments: in microfluidic chips, on microscope slides, and in benchtop bioreactors. Time-series observations will allow us to assess changes in population composition and, via microscopy, the spatial distribution of subpopulations. In addition, we will develop novel machine learning tools for model calibration and validation. Toward our long-term goal, our first aim will be to construct and validate computational models of lab-scale communities to address specific questions about spatio-temporal population dynamics. One set of experiments will focus on horizontal gene transfer. In another set of experiments, we will investigate mechanisms by which bacterial cells compete with their neighbours by delivery of toxins (a behaviour exhibited within the human gut) and by toxin delivery between cells in direct contact (mimicking interactions between plant pathogens and their competitors). These studies will employ single-cell resolution time-lapse microscopy. In a complementary aim, we will investigate interacting bacterial populations in small colonies each containing many thousands of cells. We will again record time-lapse images of population behaviour. At this scale, we will employ a modelling approach that does not track individual cells, but instead describes colony structure in terms of local population density. These models can be scaled-up to describe the dynamics of bacterial biofilms. Finally, another aim is to develop new machine learning approaches to enhance computational modelling efforts. These will improve efficiency of model calibration and validation. In addition, we will conduct experiments in benchtop bioreactors to assess and validate our previously developed deep learning algorithms for control and characterization of mixed cultures as tools for optimization of biomanufacturing processes. Trainees involved in these projects will develop interdisciplinary skills in computation, experimental, and design skills that will position them to serve as leaders in Canada's burgeoning bioeconomy.