PROJECT SUMMARY Understanding selection in complex evolving populations is a common theme across the biomedical sciences. Examples include the characterization of driver mutations that lead to cancer, pathogen evolution to escape human immune responses, and the growth of antibiotic-resistant bacteria. Recent experimental advances have substantially increased the availability of temporal genetic data, which could be exploited to detect selection with greater accuracy and precision. However, inferring selection from temporal genetic data remains technically challenging. The central goal of my research is to develop and apply efficient computational and statistical methods to quantitatively describe evolutionary dynamics, including the role of selection in evolution. Drawing on novel approaches derived from statistical physics, we will develop robust, scalable, and interpretable methods to infer the fitness effects of mutations from temporal genetic data, accounting for features such as genetic linkage, epistasis, and time-varying selection. These methods will be integrated into a software package in order to make them more widely accessible to the community. We will focus on two specific applications: 1) investigating the evolution of human immunodeficiency virus (HIV)-1 to evade adaptive immune responses, a prototypical example of rapid and complex evolution, and 2) interpreting massively parallel assays of protein function. Our research program will create new tools for understanding complex evolving populations and apply them to elucidate host-pathogen coevolutionary dynamics and to improve widely used high-throughput experimental techniques.