EVEscape is a cutting-edge AI model designed to predict viral mutations that enhance immune evasion, such as preventing antibodies from binding effectively. By integrating deep learning models with detailed biological and structural data of viruses, EVEscape accurately forecasts the evolution of viral variants.
What sets EVEscape apart is its ability to forecast new viral variants using data available at the onset of outbreaks. This capability is pivotal for proactive pandemic preparedness and designing vaccines targeting emerging variants.
Pascal Notin, a DPhil student at Oxford’s Applied and Theoretical Machine Learning group (OATML), highlighted, “Our study demonstrates EVEscape’s potential to predict prevalent mutations and concerning variants of SARS-CoV-2 accurately, had it been deployed at the pandemic’s outset.”
The study validated EVEscape’s predictions using early COVID-19 data from February 2020, accurately identifying prevalent SARS-CoV-2 mutations and anticipated changes in antibody-based therapies efficacy.
EVEscape’s applicability extends beyond COVID-19 to influenza, HIV, and emerging viruses like Lassa and Nipah, making it a versatile tool for pandemic surveillance and vaccine development.
Associate Professor Yarin Gal from Oxford’s Department of Computer Science emphasized, “EVEscape’s strength lies in leveraging initial pandemic data for precise variant predictions.”
Initially developed to predict disease-related genetic mutations like cancer, EVEscape’s unsupervised learning approach ensures robust predictions even for rare diseases.
In response to the COVID-19 crisis, EVEscape now offers real-time monitoring of SARS-CoV-2 variants, publishing biweekly updates crucial for global health organizations. This proactive approach supports the development of effective vaccines and therapies.
Researchers worldwide can access EVEscape’s complete code online, empowering them to combat current and future viral threats effectively.
