Potential Sciences launches to use simulations to speed early drug design

In short: UK startup Potential Sciences says it can use computer simulations to create chemical data that helps AI find promising drug ideas faster.

What happened

Potential Sciences, a new British startup that launched on November 5, is trying to speed up the earliest stage of drug development. It plans to generate large amounts of “synthetic” biochemical data, meaning data made by computers rather than lab experiments.

The company says it will use physics-based simulations to model how atoms and molecules interact, then scale up to amino acids and proteins. Proteins are large molecules in the body that drugs often need to attach to, like a key fitting into a lock. Potential Sciences hopes these simulations will create much bigger datasets than today’s commonly used public sources, such as the Protein Data Bank.

With those larger datasets, the startup wants to train AI systems that can better predict which drug candidates are likely to work, and which should fail early. Founder and CEO Hatem Helal told WIRED the goal is to make weak drug ideas “fail fast,” before teams spend years and large sums testing them.

Potential Sciences says it has raised seed funding in the tens of millions of dollars. Investors include SovAI, the UK government’s AI-focused venture fund, and Twin Path Ventures. The company did not share the exact amount or its valuation.

Why it matters

If prediction improves, drug makers could spend less time running expensive trial-and-error lab work at the start. That could make it more practical to pursue treatments for rare diseases, although experts note that safety testing in animals and people will still take a long time. Helal also acknowledged a risk, tools that can reduce toxicity could potentially be misused to increase it.

Source: Wired