A startup rooted in Jewish innovation and entrepreneurship is unlocking the potential of gene therapy for countless thousands of patients.
HoudiniBio is working to overcome one of the biggest hurdles in the field – making sure cells in the patient's body don’t reject the “fixed genes” that have been “edited” to replace those with a flaw.
The genes are designed to cure a wide range of conditions such as haemophilia, cystic fibrosis, dementia, inherited blindness and cancer.
But developing and producing these therapies is a hugely expensive business – A single dose for one patient can cost the NHS £3 million, largely because finding a way to design the fixed gene so that the body doesn't shut it down has, until now, been a needle-in-a-haystack endeavour.
But HoudiniBio says it has cracked that problem, using AI to accurately predict which particular design, out of an infinite number of possibilities, will actually work.
The company was co-founded by Jonathan Cohen-Gold, a PhD graduate in molecular medicine at the University of Cambridge, and a member of the city's Beth Shalom Reform Synagogue.
The science builds on a breakthrough by his supervisor, Paul Lehner, Professor of Immunology, who attends the same shul.
Lehner and his team discovered a cellular “off switch” called the Human Silencing Hub (Hush) complex, which is designed to protect DNA by shutting down genetic material that looks like it came from a virus.
But Hush is a double-edged sword as it doesn't distinguish between foreign genes that will cause harm and those that are there to help.
During his PhD research, Cohen-Gold worked on identifying the molecular rules that allow some DNA sequences to escape Hush recognition.
HoudiniBio aims to provide the "camouflage" for pharma companies that are developing gene therapies but can't get them under the radar.
Cohen-Gold is CEO of the company, which he co-founded with Lee Dunham (chief business officer) in 2025 and which takes its name from the legendary Jewish escapologist Harry Houdini, who was born Erik Weisz.
"I spent my whole PhD obsessing about ways genes could escape detection by the Hush complex and from that Houdini just sprang to mind," Cohen-Gold said.
The company, based in London, has just secured £1.5 million to fund further research and has emerged from stealth – the quiet research phase before a startup announces itself to the world.
Cohen-Gold, aged 32, says the company's vision is to overcome the commercial bottlenecks that currently make gene therapy so rare and expensive.
"There are other people in the race," he went on, "including large pharma companies.
“From the data that we're able to see from these competitors, we're able to substantially outperform them.
"And that's because we were the first to understand the HUSH complex and to discover these novel rules.
"And so this is our own proprietary IP [intellectual property]. Typically, other companies rely on a lot of trial and error.
"They get a piece of DNA and then they make thousands and thousands and thousands of different variants, just testing them empirically to find the best one."
HoudiniBio accelerates that process using an AI-powered DNA design engine that finds the right way to encode the therapeutic gene.
A single gene can have billions of possible “spellings”, so HoudiniBio finds the “spelling” that won't trigger a shutdown by the body's cells.
Pharma companies bring their initial gene therapy to HoudiniBio, to have the camouflage applied.
"We're able to look at the code of the DNA itself, and our models diagnose that sequence," continued Cohen-Gold.
"They might say here's a piece of sequence that we predict is really going to attract the Hush complex, here's the piece of sequence that we think might attract it less.
"And then the model, whilst preserving the actual drug, plays around with that sequence until it diagnoses the whole thing as being resistant to the Hush effect."
It is a game changer, he said, that will allow pharmaceutical companies to produce gene therapies at a far lower cost. Patients will need far smaller doses, which reduces any side effects, and the treatment itself will be far more effective over time.
He added that the company has demonstrated that the science works in the lab, and has just embarked on trials with mice.
Clinical trials with human patients could realistically take place in three to five years.
Cohen-Gold grew up in Gantshill, Essex and attended Ilford Jewish Primary School before his family moved to Cambridge.
He studied for his BSc at King's College, London, then "hustled" a job in venture capital after his PhD at Cambridge so he could learn how to build a tech company.
But he admits molecular genetics wasn't his first love, instead naming his first career as a songwriter, working with his twin brother Eli before he went to university.
"We signed a record deal and a global publishing deal. We worked with artists like One Direction and Ellie Goulding, and we were really involved."
To get more news, click here to sign up for our free daily newsletter.