The Scotsman

The drugs do work – thanks to AI

U David Lee and Peter Ranscombe find out how scientists in Scotland are using artificial intelligen­ce in healthcare, from designing drugs through to analysing diagnostic images

Mention the phrase “drug discovery” and it’s natural to imagine people wearing white coats and goggles, wielding pipettes and staring down microscope­s as they sit at laboratory benches. The picture is one of trial and error, incrementa­l steps, testing different chemical compounds one by one to see if they could be used to treat a disease.

But how about screeds of data being displayed on a computer screen? How about long sequences of ones and zeros being analysed to look for patterns? How about computer scientists and programmer­s rubbing shoulders with chemists and medics?

Throughout his career, Lee Cronin has been bringing together the worlds of chemistry and computing. He joined the University of Glasgow as a lecturer in 2002 and was appointed as regius professor of chemistry in 2013, aged just 39.

For one of his latest projects, he teamed up with Juan Manuel Parrilla Gutierrez, who had worked with Cronin at the University of Glasgow before lecturing in robotics at Glasgow Caledonian University. They built a machine-learning model to understand how molecules fit together, based on their shape and the electric charges on their atoms.

“Our algorithm is like Playdoh, the children’s toy,” explains Gutierrez. “It’s like adding or removing pieces of dough to see how well they fit.

“We then used a tool, which took that three-dimensiona­l (3D) volume – that Playdoh shape – and identified which molecule that 3-D shape represente­d. It turned the shape into an actual molecule.”

Cronin adds: “That model is like the ‘imaginatio­n’ – it’s what might be possible. What I’ve been doing in parallel for the past decade or so is building a ‘chemputer’, which is an automated chemistry robot.

“If you link the robot to the imaginatio­n then the robot can actually make new molecules, potential drugs for testing in labs. We’re doing this right now in my lab in Glasgow and also in my company, Chemify.”

Cronin spun Chemify out from the University of Glasgow in 2022 and last year raised $43 million (£35m) of funding, consisting of $34m from venture capital investors led by United Statesbase­d Triatomic Capital and the equivalent of $9m from the UK government’s innovation accelerato­rs programme. “There are now 80 people in the company and it’s building its first factory in Glasgow at the moment,” he says.

His team at the university is working with precision medicine specialist­s to develop drugs to treat colon cancer, as well as looking at fields including malaria. Cronin also won funding from the US National Institutes of Health to help solve the opioid addiction crisis.

“Opiates are very good for treating pain, but they’re addictive,” he explains. “Can you change the shape of the molecule to basically hit the pain receptor but not make it as addictive – that’s what we’re trying to do.”

Cronin praises the University of Glasgow for supporting his aim to build a multi-disciplina­ry team, which brought together chemists with computer scientists such as Gutierrez. He wants universiti­es to work together with Scottish Enterprise and the Scottish Government to build more infrastruc­ture, including science parks.

“Companies will move here because Scotland is a wonderful place to be,” adds Cronin, who moved from England to work in Scotland. “There are lots of good universiti­es, there’s lots of talent, there’s just a massive lack of ambition and a massive lack of planning.”

Another Scottish company that is using artificial intelligen­ce (AI) in drug discovery is Exscientia, which was spun out from the University of Dundee in 2012 and opened its head office in Oxford in 2018. The firm raised $510m (£407m) in 2021 when it listed its shares on the Nasdaq stock market in New York, the largest initial public offering (IPO) by a European biotechnol­ogy company, and has now grown to employ around 450 people in six countries, with about 20 of those members of staff still based in Dundee.

Exscientia became the first company to design a molecule using AI that went into clinical trials as a drug candidate, and now has four molecules undergoing testing in clinics, tackling conditions in areas ranging from psychiatry through to the immune system. As well as partnering with industry giants including Bristol Myers Squibb and Sumitomo Pharma, the company also has its own internal focus on cancer.

Using AI has allowed the company to cut the time it takes to go from having an idea to identifyin­g a potential compound for a drug from four-and-a-half years down to between 12 and 18 months. The number of molecules

There’s lots of talent here – but a massive lack of ambition and a massive lack of planning Lee Cronin

that need to be examined during that process has dropped from 2,500 to between 250 and 400, slashing the associated costs by about 80 per cent.

“I can guarantee that within the next ten years you’ll start to see drugs on the market that were invented or discovered using AI,” says Dave Hallett, the company’s chief scientific officer and interim chief executive. “I truly believe the suite of tools we call AI will have made a positive contributi­on to all aspects of drug discovery and developmen­t within the next ten years, whether it’s a small molecule that you might take as a tablet like paracetamo­l, or antibodies, or cell therapies.”