Isomorphic Labs: Alphabet's AI Drug Discovery Venture

AI's Impact on Drug Discovery: The Launch of Isomorphic Labs

Artificial intelligence is dramatically reshaping the landscape of drug discovery. Numerous organizations are leveraging AI to transform a significant practical challenge into a manageable information-based problem.

Alphabet's New Venture: Isomorphic Labs

Alphabet, Google’s parent company, has recently entered this arena with the establishment of Isomorphic Labs. The initiative is spearheaded by Demis Hassabis, who also leads DeepMind.

Details surrounding the company’s launch were limited to an initial blog post and a general FAQ. Isomorphic Labs aims to “develop a computational platform for understanding biological systems from fundamental principles, ultimately leading to innovative disease treatments.”

Underlying Assumptions and Challenges

This founding statement rests on several key assumptions. A primary one is the feasibility of computationally modeling biological systems in a way that accelerates drug discovery.

Over the past five years, substantial investments – totaling hundreds of millions of dollars – have been directed towards companies pursuing similar objectives. However, a breakthrough AI-discovered drug for a previously incurable condition has yet to emerge.

The reasons for this lack of immediate success are complex and extend beyond the scope of this discussion. It’s important to recognize that AI systems are not instant solutions, but rather integral components within a lengthy and resource-intensive process that still relies heavily on traditional laboratory methods.

A Balanced Perspective from Demis Hassabis

Hassabis acknowledges the complexity, describing biology as “an information processing system, albeit an extraordinarily complex and dynamic one.”

He quickly adds nuance, explaining that the company’s name, Isomorphic Systems, reflects the idea that information and biological systems may share a fundamental structure. The term isomorphic signifies similarity in form despite differing origins.

Leveraging AlphaFold's Success

This reasoning is likely influenced by the remarkable success of DeepMind’s AlphaFold. This AI-powered system revolutionized protein folding prediction, establishing a new standard in a notoriously challenging field.

DeepMind’s learning systems demonstrate a strong capacity for generality and knowledge transfer. This means their underlying structure can be adapted to diverse tasks.

If, as AlphaFold suggests, biological systems are well-suited for this type of simulation and analysis, Hassabis’s optimistic assessment of Isomorphic Labs’ potential may prove accurate.

Looking Ahead: A Long-Term Endeavor

However, tangible results are not expected immediately. Despite the advantage of DeepMind’s existing AI research – which will remain distinct but potentially shared – Isomorphic Labs is essentially beginning its work from the ground up.

The company is currently assembling a “world-class multidisciplinary team.” It may take a year or two before the first indications of progress emerge from this ambitious undertaking.