Researchers have developed a new artificial intelligence (AI) tool which they say could help predict new viral variants before they emerge.
The model, named EVEscape, predicts the likelihood that a viral mutation will enable it to escape immune responses -- for instance, by preventing antibodies from binding. EVEscape was developed by scientists at the University of Oxford and Harvard Medical School who say it could help in the design of vaccines to target variants of concern before they become prevalent.
“Our study shows that had EVEscape been deployed at the start of the Covid-19 pandemic, it would have accurately predicted the most frequent mutations and the most concerning variants for SARS-CoV-2,” said Pascal Notin, co-lead author of the research paper published in Nature.
The model was developed by the Debora Marks Lab at Harvard Medical School and Oxford Applied and Theoretical Machine Learning Group (OATML) at the University of Oxford. The core component of EVEscape is EVE -- short for “evolutionary model of variant effect” -- which the research team originally developed to predict the effects of genetic mutations on human disease risk. EVEscape combines a deep-learning model of how a virus evolves with detailed biological and structural information to enable predictions about the variants most likely to occur.
The model of protein sequences helps researchers understand which mutations preserve the fitness of a given virus. Researchers say that it quantifies the “viral escape potential of mutations at scale” and can be applied before surveillance sequencing, experimental scans, or three-dimensional structures of antibody complexes are available.
For the study, they tested the model using information only available at the beginning of the COVID-19 pandemic in February 2020. It successfully predicted which SARS-CoV-2 mutations would occur during the pandemic and which would become the most prevalent, according to researchers.
“We demonstrate that EVEscape, trained on sequences available before 2020, is as accurate as high-throughput experimental scans at anticipating pandemic variation for SARS-CoV-2,” the authors wrote.
The model also predicted which antibody-based therapies would lose their efficacy as the pandemic progressed; the virus developed mutations to escape these treatments.
In principle, this model could also be applied to other viruses including influenza, HIV, and understudied viruses with pandemic potential such as Lassa and Nipah.
A key advantage of the model, say researchers, is that it would already be available at the start of a pandemic, so there would be no need to wait for relevant antibodies to arise in the population to predict which variants were the most concerning.
“We want to know if we can anticipate the variation in viruses and forecast new variants -- because if we can, that’s going to be extremely important for designing vaccines and therapies,” said Debora Marks, a co-author and professor of systems biology in the Blavatnik Institute at Harvard Medical School.