Researchers at the University of Cambridge have used a machine learning model to design a vaccine targeting the entire sarbecovirus family, which includes human and animal coronaviruses. Published in the Journal of Infection, the study marks the first time an AI-designed vaccine has been tested in humans, offering a method to preemptively protect against future viral mutations.
Predicting immutable targets
Traditional vaccines require constant updates as viruses mutate, leading to seasonal flu shots and repeated COVID-19 boosters. Machine learning models change this dynamic by scanning genetic data from thousands of related viruses to predict which components are most likely to remain stable over time. Targeting these immutable regions allows a single vaccine to theoretically protect against an entire family of viruses and their future variants.
Trial results and limitations
The Cambridge team applied their model to the sarbecovirus family. The AI identified stable viral components, and researchers then narrowed these down through careful in vivo experimentation to create a super-antigen for the vaccine. Initial human trials involving 39 participants showed the vaccine successfully activated the immune system to produce virus-fighting antibodies. The researchers reported it was well tolerated across all four doses with no significant safety concerns.
However, the trial was small, the immune response was modest, and the duration of protection remains unclear. Further testing is required to validate the efficacy of the approach in larger populations.
DNA-based delivery advantages
The developed vaccine relies on DNA rather than mRNA. This design eliminates the need for needle injections and increases stability, allowing for easier transport and storage without the ultra-cold conditions required for mRNA vaccines. Scientists noted this stability could aid in addressing outbreaks like the current Ebola crisis in the Democratic Republic of Congo, where doctors lack a vaccine for the new strain.
Why this matters for science and research professionals
For those tracking AI for Science & Research, this study demonstrates how machine learning can shift vaccine development from a reactive process to a predictive one. By identifying stable genetic markers across viral families, research teams can design broader-spectrum therapeutics before a pathogen jumps to humans.
"We've converted vaccine development from being reactive to being future proof," study coauthor Jonathan Heeney said in a statement. "Our vaccines will continue to provide protection against viruses even as they mutate into new strains. We've overcome the problem of traditional vaccines, which have limited protection. It means we can escape the constant cycle of chasing the virus variants circulating in humans and updating the vaccines to try to catch up, like a dog chasing its tail."
Your membership also unlocks:
