From WIRED · Joel Khalili · · 2m
AI startup aims to speed up drug discovery
Potential Sciences uses physics simulations to build datasets that could train AI to identify drugs faster, bypassing costly lab work.
In brief
Potential Sciences uses physics simulations to build datasets that could train AI to identify drugs faster, bypassing costly lab work. British startup Potential Sciences launched with tens of millions in funding to use AI and physics simulations to accelerate drug discovery and reduce experimentation costs. Originally…
New startup takes on drug design
British startup Potential Sciences believes AI can speed up the slow process of drug design by using physics simulations to build chemical datasets instead of lab work.
How the approach works
Potential Sciences plans to simulate interactions between atoms, molecules, amino acids and proteins to create large-scale datasets, then train AI models to identify targeted drugs for specific diseases.
Founder's vision for the company
CEO Hatem Helal says the company wants molecules that won't work in the clinic to fail fast by having better predictors of how they interact with human biology.
“We want molecules that are never going to work in the clinic to fail fast.”
Funding and launch details
Potential Sciences launched on November five with seed funding in the tens of millions of dollars from SovAI, the UK government's AI venture capital fund, and Twin Path Ventures.
Potential impact on rare diseases
By reducing experimentation costs, Helal says the approach could make it economical for pharmaceutical companies to pursue drugs for rare conditions that have previously gone untreated.
The scale of the challenge
Helal notes there are around ten to the power of sixty ways to assemble atoms into molecules for humans. Drug discovery is a search problem about navigating that space into promising molecules.
Expert skepticism on predictions
Peter Wolynes, a chemistry professor at Rice University, says protein engineering is currently at the level of the Romans, not the Renaissance, in terms of advancement.
“the engineering is at the level of the Romans, not of the Renaissance, in terms of how protein engineering is done”
Testing remains the bottleneck
Wolynes says identifying promising drug candidates is not the rate-limiting step. The cost of testing safety and efficacy on mice, patients and normal people far exceeds design costs.
“The cost of designing a drug is nothing compared to figuring out what it will do to mice; to really sick patients; to normal people.”
Dual-use risk and unanswered questions
Helal acknowledges that tools to minimize toxicity could be flipped to maximize it. Potential Sciences has not yet settled on revenue models or whether to make models freely available.
“If we can design tools that perfectly minimize toxicity, it's a bit flip to maximize it as well.”
What matters most
British startup Potential Sciences launched with tens of millions in funding to use AI and physics simulations to accelerate drug discovery and reduce experimentation costs.
