Effective vaccines played a crucial role in mitigating the impact of the COVID-19 pandemic, saving millions of lives and reducing disease severity. Five years later, however, SARS-CoV-2 continues to circulate and evolve into new variants, necessitating the continuous development of updated vaccines. The delay in designing, manufacturing, and distributing these vaccines raises a significant challenge: How can scientists prepare vaccines for viral versions that haven’t yet emerged? A promising solution has been developed by a team of scientists from Harvard Medical School, the Massachusetts Consortium on Pathogen Readiness (MassCPR), and other institutions. They created an AI model called EVE-Vax, which combines evolutionary, biological, and structural data to predict and design viral surface proteins that could appear as the pathogen mutates. This model, described in a May 8, 2023, issue of the journal Immunity, has shown remarkable potential in forecasting the future evolution of viruses and assisting in vaccine design. The foundation of EVE-Vax lies in the earlier work of study co-senior author Debora Marks, a professor of systems biology at Harvard Medical School. More than a decade ago, Marks and her team began investigating whether they could leverage millions of years of evolutionary genetic data to predict the structure and function of proteins. Their initial AI model, EVE (Evolutionary Model of Variant Effect), successfully interpreted human gene variants as either benign or disease-causing, using large-scale evolutionary data. During the COVID-19 pandemic, the team adapted EVE to focus on viral behavior, resulting in EVEscape. This model accurately predicted the most frequent SARS-CoV-2 mutations and identified the variants of greatest concern, demonstrating its potential for early detection of highly infectious strains. The success of EVEscape prompted the researchers to further explore whether their model could forecast the future evolution of rapidly mutating viruses like SARS-CoV-2, which are typically updated annually based on educated guesses made up to a year in advance. To address this challenge, the researchers developed EVE-Vax. Using the model, they designed 83 novel versions of the SARS-CoV-2 spike protein, each with a unique combination of up to ten mutations. spike protein is the primary means by which the virus infects human cells. To validate the effectiveness of the AI-designed proteins, the team collaborated with experimental researchers, including Jeremy Luban from UMass Chan Medical School, Jacob Lemieux from Massachusetts General Hospital, and Michael Seaman from Beth Israel Deaconess Medical Center. They conducted lab experiments using safe, nonreplicating versions of SARS-CoV-2. The results confirmed that the viruses with "designer" spike proteins infected human cells and elicited immune responses that closely mirrored real-life immune reactions at various stages of the pandemic. According to Marks, the key insight behind EVE-Vax is that evolutionary data can reveal the potential pathways of viral evolution, allowing scientists to anticipate future mutations and design vaccines accordingly. The model’s ability to generate hundreds of new spike proteins quickly and cost-effectively is a significant advantage. Traditional vaccine design relies on a variety of methods, but EVE-Vax offers a novel and potentially more accurate approach. For instance, the researchers demonstrated that EVE-Vax could have predicted the extensive immune escape from the omicron-targeted booster vaccine, providing valuable knowledge that could have guided the development of a more effective booster. "With EVE-Vax, we can predict the immune response, not just the mutations, which is much more useful in practical scenarios," Marks explained. The team is now expanding the applications of EVE-Vax to other rapidly evolving viruses, such as avian influenza, which poses an increasing threat globally. One notable benefit of the model is its capability to work with limited data, making it particularly valuable for understudied viruses like Lassa and Nipah, as well as emerging pathogens. By providing scientists with a comprehensive view of likely evolutionary paths, EVE-Vax aims to enhance vaccine preparedness and effectiveness. Industry insiders and experts have expressed enthusiasm about the potential of EVE-Vax. The model's integration of multiple data sources and its ability to generate realistic protein designs are seen as groundbreaking advancements in the field of vaccine design. Companies and research institutions involved in virology and immunology are closely monitoring the progress of EVE-Vax, recognizing its potential to streamline and improve the process of creating vaccines for highly mutable viruses. EVax represents a significant step forward in leveraging AI to protect public health, and its continued development and application could have far-reaching implications for future pandemic preparedness.