British startup Potential Sciences is betting that artificial intelligence and physics-based simulations can make drug discovery faster, less expensive and more precise. The company plans to build large chemical datasets by modeling interactions between atoms, small molecules, amino acids and proteins, then use that data to train AI systems that can predict targeted drugs for specific diseases before costly laboratory experiments begin.
Using physics to guide drug discovery
Potential Sciences aims to use fundamental principles of physics to simulate how chemical compounds interact at increasingly complex levels. The resulting datasets could help its predictive models identify drug candidates likely to interact with disease-related proteins.
“We want molecules that will never work in the clinic to fail quickly,” Potential Sciences founder and CEO Hatem Helal told WIRED. “That means we can more accurately predict how they will behave when they interact with human biology.”
Founded on November 5, the startup said it has secured tens of millions of dollars in seed funding from investors including SovAI, the UK government’s AI venture capital fund, and Twin Path Ventures. The founders declined to disclose the exact amount raised or the company’s valuation.
Aiming for more personalized treatments
If the approach works as intended, it could accelerate the development of treatments for a wide range of diseases. Helal said reducing experimental costs could make it economically viable for drug companies to pursue therapies for rare conditions that have previously received little or no treatment research.
“We want to enable more tailored treatments for diseases, the idea being that medicines can be one-off combinations of molecules tailored to a person’s unique biology,” he said.
Challenges in predicting drug performance
The central question is whether Potential Sciences can accurately predict how well its drug candidates will bind to disease-related proteins. Achieving that will require significant progress in protein engineering and computational modeling.
“I’m confident it will be an improvement,” said Peter Wallins, a chemistry professor at Rice University. But, he added, “engineering is at the level of the Romans, not the Renaissance, in terms of how protein engineering is done.”
Industry figures including Demis Hassabis, who won the Nobel Prize in Chemistry for his work with Google DeepMind, have said for a long time that advances in AI could eventually drive major scientific breakthroughs, just as technology has recently expanded what is possible in mathematics. Startups such as Potential Sciences hope drug discovery will be among the first scientific fields transformed by these advances.
Source: www.wired.com