Finding New Drugs Is Laborious. This AI Startup Is Skipping The Lab Work

Potential Sciences, an AI startup, aims to revolutionize drug discovery by bypassing traditional lab work. The company plans to build extensive chemical datasets by simulating atom interactions using the principles of physics, followed by smaller molecules, amino acids, and proteins. With this data, Potential Sciences intends to train AI models and prediction systems capable of identifying highly targeted drugs for specific diseases, potentially making it more cost-effective for pharmaceutical companies to develop treatments for rare conditions.

Founder and CEO Hatem Helal emphasizes the importance of quickly identifying molecules that are unlikely to succeed in clinical settings, using the vast potential of synthetic data generated through simulations. If successful, this approach could transform drug discovery into a more efficient and affordable process, paving the way for bespoke treatments tailored to individual human biology.

Despite the optimism surrounding the technology, experts like Peter Wolynes, a professor of chemistry at Rice University, caution that the engineering behind protein engineering is still at an early stage, and predicting how well drug candidates will bind to disease-related proteins remains a significant challenge. The startup's approach differs from other AI drug discovery companies, as it emphasizes simulating chemistry rather than relying on a single public database of 3D atomic structures.

While AI drug discovery is a competitive field, Potential Sciences may benefit from the UK government's investment through its newly established SovAI fund. However, challenges remain, including the long and costly process of testing drug safety and efficacy, as well as the potential misuse of technology to create highly dangerous chemicals.