A new AI tool developed by researchers at Harvard Medical School (HMS) and the University of Oxford called EVEscape has the ability to predict viral variants before they even emerge, greatly benefiting vaccine design. The tool consists of two elements: one that models evolutionary sequences to anticipate changes in the virus, and another that provides detailed biological and structural information.
In a study published in Nature, the team behind EVEscape revealed that if the tool had been in use at the beginning of the COVID-19 pandemic, it would have accurately predicted the most frequent mutations in the SAR-CoV-2 virus, as well as the emergence of concerning variants. Remarkably, the tool also demonstrated accurate predictions for other viruses, including HIV and influenza.
With the proven capabilities of EVEscape, the researchers are now using it to look ahead and predict SARS-CoV-2 variants of concern. They are releasing rankings of these variants every two weeks, offering valuable insights for vaccine and therapy design.
EVEscape is built upon a previous AI model called EVE (evolutionary model of variant effect), which was initially developed to predict the functionality of proteins based on evolutionary data across species. This earlier model helped identify genetic mutations associated with human diseases. During the COVID-19 pandemic, the researchers repurposed EVE to predict viral variants. By incorporating biological and structural details of the virus, including information about regions targeted by the immune system, they enhanced EVE’s capabilities for virus prediction.
To evaluate EVEscape’s effectiveness, the team used data from January 2020 to predict the future behavior of the SARS-CoV-2 virus. The results showed that EVEscape’s predictions of SARS-CoV-2 mutations during the pandemic were as accurate as experimental approaches that directly test the virus’s ability to bind to antibodies. Moreover, EVEscape outperformed experimental methods in predicting the most prevalent mutations. Importantly, the tool provided faster and more efficient predictions compared to lab-based testing, as it did not rely on waiting for antibodies to become available for testing. EVEscape also predicted the loss of effectiveness of certain antibody-based therapies as new variants emerged.
The success of EVEscape does not confine itself solely to SARS-CoV-2. The researchers demonstrated its applicability as a predictive tool for HIV and influenza as well.
The code for EVEscape is openly available to researchers on GitHub, along with bi-weekly rankings of the most concerning SARS-CoV-2 variants on their website. The team also shares these rankings with public health entities, including the World Health Organization.
In conclusion, the development of EVEscape has ushered in a new era of predictive tools for viral variants. By accurately anticipating these variants, scientists can better design vaccines and therapies, ultimately combating the spread of infectious diseases more effectively. The availability of EVEscape’s code and rankings will enhance collaboration among researchers and aid public health organizations in staying ahead of viral mutations.
