British startup Potential Sciences aims to bypass traditional lab experimentation by using physics-based simulations to train artificial intelligence models for drug design. The company plans to build large-scale chemical datasets by simulating interactions between atoms, small molecules, and complex proteins to identify targeted treatments for specific diseases. This approach seeks to reduce the costs of experimentation, potentially making it economically viable for pharmaceutical companies to develop medicines for rare conditions.
Founder and CEO Hatem Helal stated that the company intends to use these predictive models to ensure that molecules unlikely to work in clinical settings fail early in the process. "We want molecules that are never going to work in the clinic to fail fast," Helal told WIRED.
The startup launched on November 5 with seed funding in the tens of millions of dollars from Twin Path Ventures and SovAI, the UK government’s AI venture capital fund. While the founders declined to disclose the exact valuation or funding figure, the investment places Potential Sciences alongside other AI drug discovery firms like Isomorphic Labs and Basecamp Research.
Potential Sciences' strategy addresses a limitation in current AI drug discovery, where many labs rely on the Protein Data Bank. Helal argued that this public database of 3D atomic structures is too small to train models advanced enough to reliably accelerate the search for viable drug candidates. To solve this, the company is constructing larger datasets using synthetic data produced through chemical compound simulations.
Academic experts expressed cautious optimism regarding the simulation-based method. Alan Fersht, an emeritus professor of chemistry at the University of Cambridge, said that accurately calculating molecular interactions and structural fluctuations could revolutionize drug development. However, Peter Wolynes, a professor of chemistry at Rice University, noted that current protein engineering technology remains in an early stage of development.
If successful, the technology could enable the creation of bespoke treatments tailored to an individual's unique biology. The company's goal is to navigate the vast chemical space of possible molecular combinations to find promising candidates for specific diseases more efficiently than current methods allow.
