From predicting influenza virus evolution to enhancing vaccine antigen design
Prediction of virus evolution is crucial for pandemic and epidemic preparedness. In this talk, we will introduce a novel framework, beth-1, to forecast the genetic evolution of the influenza virus. Unlike traditional phylogenetic methods, this method models evolution from a site-based perspective, learning the fitness dynamics of mutations from global genomic and epidemiological data. Through retrospective, prospective, and experimental validations, we demonstrated that the proposed method offers more accurate predictions of future virus populations compared to existing approaches. To assess the public health impact of vaccine strains, we also developed the VE-GD model to estimate vaccine effectiveness by analyzing virus genome. This model bridges molecular variations and population-level vaccine protection, achieving 87% prediction accuracy for influenza vaccines and 95% for COVID-19 vaccines. The talk will also explore new insights into the dynamic changes in virus populations.
In-person event. If you wish to join online, please register here.
