UC San Diego

Quantifying the evolutionary dynamics of beneficial mutations in microbial populations relies on a massive sampling of nascent mutations, which has recently become possible by innovative barcode techniques. Despite barcode lineage tracking (BLT) experiments offer a straightforward means of capturing mutation frequency trajectories, accurately quantifying their dynamics remains a challenge. This dissertation addresses the critical issue of precisely and accurately estimating the fitness effects, mutation rates, and the distribution of fitness effects (DFE) associated with beneficial mutations identified in BLT experiments. First, I investigate the limitations of existing methods used to infer fitness effects and identify potential reasons for failure. To address this issue, I develop a novel Bayesian filtering method for fitness inference, which significantly enhances the accuracy of estimation compared to current approaches.Secondly, to estimate the rate of beneficial mutations within specific classes, I develop a stochastic model of mutant population dynamics. By applying this model, I infer the rate of beneficial mutations in the BLT experiment, providing valuable insights into mutation dynamics. Lastly, I highlight an inherent bias in our observations of beneficial mutations through adaptive evolution relative to the true underlying DFE. To address this disparity, I propose a method for inferring the mutation rate spectrum by back-calculating the observed DFE. The work presented in this dissertation establishes a statistically solid foundation for the analysis of the BLT experiment.