We're targeting something the virus can't change easily.
In a Cambridge laboratory, scientists and algorithms have joined forces to ask an ancient question in a new way: not what divides our enemies, but what holds them together. By training machine learning on the conserved genetic architecture shared across entire virus families, researchers at the University of Cambridge and DIOSynVax have produced a vaccine candidate that may protect against coronaviruses not yet born into human populations — a first human trial of 49 volunteers suggesting both safety and broad immune response. The ambition is nothing less than a reorientation of public health from perpetual crisis to quiet foresight, from chasing what has already arrived to standing ready for what has not.
- Humanity has long fought pandemics in arrears — each new outbreak forcing a desperate sprint to design, test, and deploy a vaccine while the virus races ahead; this technology proposes to end that chase entirely.
- A needle-free microfluid jet delivered a single 'super-antigen' vaccine to 49 healthy volunteers, triggering immune responses against SARS-CoV-2, the original SARS, and bat coronaviruses that have never yet crossed into humans.
- The same AI framework is now being aimed at Ebola — where the Bundibugyo strain is actively spreading across the Democratic Republic of the Congo and Uganda — and at bird flu strains circulating across continents and contaminating food supplies.
- A second trial phase enrolling more than 200 participants is underway, and researchers argue that clinically validated universal vaccines, deployed before an outbreak begins, could spare millions of lives and prevent the economic devastation of lockdowns.
- The deeper disruption is philosophical: public health built on anticipation rather than reaction would require institutions, funding bodies, and governments to invest in threats that have not yet materialized — a profound shift in how societies value prevention.
In a Cambridge laboratory, researchers have begun designing vaccines that do not wait for a pandemic to arrive. Rather than targeting a specific pathogen after it emerges, the approach uses machine learning to identify what remains constant across entire virus families — the conserved features a virus cannot easily mutate without losing its ability to survive. These stable elements become the target, a so-called "super-antigen" capable of priming the immune system against variants that may not yet exist in human populations.
Professor Jonathan Heeney of Cambridge's veterinary medicine department described the logic as an inversion of traditional vaccine science: instead of reacting to what a virus has become, you ask what it cannot change. Researchers fed machine learning algorithms every catalogued coronavirus genetic sequence from surveillance programs worldwide, then applied structural biology to isolate the essential, unchanging core. The result is a vaccine designed not for one outbreak, but for an entire family of potential ones.
The first human trial enrolled 49 volunteers at sites in Cambridge and Southampton, delivering the vaccine through a needle-free microfluid jet — a high-pressure liquid stream that carries genetic blueprints directly into skin cells. Results published in the Journal of Infection confirmed the vaccine was safe and generated immune responses against SARS-CoV-2, the original SARS virus, and related bat coronaviruses not yet known to infect humans. A second phase, enrolling more than 200 participants, is now underway.
The implications reach well beyond coronavirus. The same framework is being applied to Ebola, where distinct strains have driven separate outbreaks — including the ongoing spread of the Bundibugyo strain in the Democratic Republic of the Congo and Uganda. Bird flu, now circulating across most continents and increasingly affecting mammals and humans, represents another urgent target. "It's about making sure that our technology can get whatever is going to pop up," Heeney said.
Professor Saul Faust of the University of Southampton, who led the initial trial, described the shift plainly: traditional vaccines often mismatch circulating viruses because pathogens evolve continuously between design and deployment. Universal vaccines, future-proofed against families of threats rather than individual strains, could change that calculus entirely — potentially saving millions of lives and sparing economies the devastation of pandemic shutdowns, if they can be validated before the next crisis begins.
In a Cambridge laboratory, researchers have begun work on a vaccine that could fundamentally alter how humanity responds to viral threats. Rather than waiting for a new pathogen to emerge, sicken thousands, and force scientists into a desperate race against time, this approach aims to get ahead of the curve entirely—designing a single shot that protects against not just one virus, but entire families of them, including variants that haven't yet jumped to humans.
The technology relies on artificial intelligence to do something counterintuitive: instead of studying what makes viruses different, it hunts for what they have in common. Researchers at the University of Cambridge and the biotechnology firm DIOSynVax fed machine learning algorithms every available genetic sequence of coronaviruses ever catalogued by surveillance programs worldwide. The system then identified the features that remain constant across all these variants—the parts of the virus that cannot easily mutate without destroying the pathogen's ability to survive. These conserved elements become the target of the vaccine, what scientists call a "super-antigen."
Professor Jonathan Heeney of Cambridge's veterinary medicine department explained the logic plainly: you gather all the genomic data from past outbreaks and current ones, apply structural biology, and ask a simple question—what stays the same, and what is essential for this virus family to exist? "We're targeting something in a virus family, which the virus can't change easily," he said. This inverts the traditional vaccine paradigm, which has always been reactive, waiting for a threat to materialize before designing a response.
The first human trial enrolled 49 healthy volunteers between 18 and 50 years old at sites in Cambridge and Southampton. The vaccine was delivered using a needle-free microfluid jet—a thin, high-pressure stream of liquid that pushes genetic blueprints directly into skin cells. Results published in the Journal of Infection showed the vaccine was safe and triggered immune responses not only to SARS-CoV-2 and the original SARS virus, but also to related coronaviruses carried by bats that could theoretically jump to human populations. Animal studies had already demonstrated the approach sparked robust protection across a range of coronavirus strains.
These early results have prompted expansion to a second phase involving more than 200 participants. The implications, researchers argue, extend far beyond coronavirus. The same machine learning framework could be applied to Ebola, where multiple strains within the same viral family have caused distinct outbreaks—the devastating West African epidemic from 2013 to 2016, and the current outbreak in the Democratic Republic of the Congo and Uganda caused by the Bundibugyo strain. "We're behind the curve, and these viruses belong to the same family," Heeney noted. A single universal Ebola vaccine could theoretically protect against all of them.
Bird flu represents another urgent target. The virus now circulates across most continents, affecting not just birds but mammals and humans, and has even contaminated the food supply in America through milk. Different clades exist, particularly in Southeast Asia, where the virus has killed people in double digits. "It's about making sure that our technology can get whatever is going to pop up and protect us, and to get ahead of that curve, instead of chasing it," Heeney said.
Professor Saul Faust of the University of Southampton, who led the initial trial phase, framed the shift in stark terms. Traditional vaccines, by the time they reach populations, often poorly match the circulating virus because pathogens evolve continuously. "This new class of universal vaccines are future-proofed," he said. "They not only protect against many variants simultaneously, but potentially against related viruses that haven't yet emerged and spilt over to humans." If such vaccines could be developed and clinically validated before an outbreak begins, millions of lives could be saved, lockdowns avoided, and economies spared the devastation of pandemic shutdowns.
What makes this a genuine paradigm shift is the move from perpetual reaction to anticipation. For decades, public health has operated in crisis mode—detecting a new threat, scrambling to understand it, manufacturing a response, and hoping to deploy it before the virus mutates beyond recognition. This technology proposes something different: identify the unchangeable core of a viral family and inoculate against it before the next crisis arrives. The first trial suggests it works. What happens next will determine whether pandemic response can finally move from chasing viruses to staying ahead of them.
Notable Quotes
This is about making one vaccine that will get them all based on their relationships. We're targeting something in a virus family, which the virus can't change easily.— Professor Jonathan Heeney, University of Cambridge
This new class of universal vaccines are future-proofed. They not only protect against many variants simultaneously, but potentially against related viruses that haven't yet emerged and spilt over to humans.— Professor Saul Faust, University of Southampton
The Hearth Conversation Another angle on the story
Why does it matter that the vaccine targets what viruses can't change, rather than what makes them different?
Because viruses evolve by mutating their surface proteins—that's how they evade immunity. But they can't mutate the parts that keep them alive. If you target those essential features, the virus has nowhere to hide. It's the difference between playing defense and playing the game itself.
So this is genuinely new? We haven't tried this before?
Not at this scale, not with this precision. Traditional vaccines chase specific strains after they emerge. This uses AI to map what's conserved across entire virus families before new variants appear. It's the difference between reacting to a fire and building firebreaks.
The trial was only 49 people. How confident should we be?
It's a proof of concept—enough to show safety and immune response. The expanded trial with 200+ people will tell us if that holds at scale. But the animal data was strong, and the logic is sound. This isn't a wild bet; it's a methodical next step.
What happens if the virus does mutate the parts you're targeting?
Theoretically, it can't—not without losing the ability to infect cells. That's the whole point. But viruses are clever. We'll learn more as we test this against real-world variants over time.
Why is Heeney so focused on Ebola and bird flu?
Because he's seen outbreaks firsthand. Ebola keeps returning in different forms from the same family. Bird flu is spreading globally right now, in multiple strains, and we're always one mutation away from a human pandemic. These aren't hypothetical threats—they're happening now, and we're still playing catch-up.
If this works, what changes?
Everything. You stop waiting for pandemics to happen. You vaccinate populations against virus families before spillover events occur. You prevent lockdowns, deaths, economic collapse. You move from reactive medicine to preventive medicine at a scale we've never managed before.