Respiratory viruses impose a significant public health burden, in large part due to their rapid evolution and ability to evade host immunity. Influenza, COVID-19, and now RSV infection, are all vaccine-preventable diseases, yet our current system of genomic surveillance is limited by significant knowledge gaps in forecasting new antigenic drift mutations and remarkably little attention to determinants of within-host and population-level viral fitness outside of antigenic sites. The objective of this project is to combine large-scale genomic surveillance and functional genomics to define determinants of epidemic success in three respiratory viruses. This project will take an integrated bidirectional approach, in which genomic surveillance is used to identify strains and mutations for experimental analysis, and in which functional genomics data are used to improve population-level inference of adaptive viral evolution. We will develop a Bayesian model to identify SARS-CoV-2 mutations positively selected within hosts and related these to ongoing surveillance data. For RSV, we will use mutational antigenic profiling of the fusion (F) protein to define and anticipate antigenic drift. We will take a population-scale approach to influenza virus by applying phylodynamic models to regional whole genome surveillance data to identify strains and mutations conferring a population-level fitness advantage. While different approaches are taken for each virus, they are universally applicable and we expect that together they will provide a foundation for more accurate prediction of emerging strains and the development of more protective vaccines.