London-based Isomorphic Labs Ltd., a drug discovery-focused artificial intelligence startup that was spun out of Google LLC’s DeepMind unit just over two years ago, today announced key partnerships with two of the world’s biggest pharmaceutical companies – Eli Lilly & Co. and Novartis AG.

The newly announced partnerships will see Isomorphic use its prowess in AI to help discover new medications that can potentially treat various diseases, the startup said.

The deals are said to have a combined value of close to $3 billion. The startup will receive upfront payment of $45 million from Eli Lilly, with as much as $1.7 billion to be paid according to performance-based milestones. As for Novartis, it has agreed to pay $37.5 million upfront, and an additional $1.2 billion in additional incentives going forward.

Isomorphic Labs was founded in 2021 by its Chief Executive Demis Hassabis under the umbrella of DeepMind, and is the developer of a technology called AlphaFold 2, which is an AI model that can be used to predict the structure of proteins in the human body. By uncovering these protein structures, researchers can identify possible new target pathways for drugs that fight a range of diseases.

The most up-to-date version of AlphaFold, announced in October, can make predictions for almost every molecule listed in the Protein Data Bank, which is an open-source database of biological molecules. It can also predict the structure of ligands, which are molecules that can be bound to receptor proteins and change the way cells communicate with one another, as well as nucleic acids, which are molecules that contain key genetic information.

Proteins, the building blocks of life, are complex molecules that twist and fold into various shapes. The structures into which a protein configures itself directly influence its behavior. As a result, studying those structures is a major priority for scientists.

Manually determining the shape of a protein can require years of work in some cases. As a result, scientists have long sought to develop software that can automate the task. Three years ago, DeepMind became the first to achieve that goal with its AlphaFold model, which demonstrated the ability to predict the structure of proteins in a matter of days.

A recent article in the journal Nature shows that the AlphaFold 2 model isn’t 100% reliable and that it sometimes makes blatant mistakes. As such, researchers use its suggestions more as a hypothesis generator, as opposed to experimental data. However, its advantage is that it can generate reasonably accurate predictions of proteins much faster than previously used methods.

In their experiments, Isomorphic’s researchers recently used AlphaFold 2 to design and synthesize a new drug for hepatocellular carcinoma, the most common kind of liver cancer. Moreover, DeepMind is collaborating with the Drugs for Neglected Diseases Initiative nonprofit organization to use AlphaFold 2 to formulate medicines for Chagas disease and Leishmaniasis, which are two rare but deadly diseases mostly found in developing nations.

Hassabis said in a statement that the companies’ shared focus on advancing groundbreaking approaches to drug design and their appreciation of “state-of-the-art science makes these partnerships particularly compelling.”

The partnerships should also relieve some of the strain off of Isomorphic Labs, which recorded a $2.4 million loss in 2021 and is reportedly under pressure to start delivering a profit.

Fiona Marshall, Novartis’ president of biomedical research said AlphaFold has the potential to transform drug discovery and accelerate the delivery of new, life-changing medicines to patients. “The collaboration harnesses our companies’ unique strengths, from AI and data science to medicinal chemistry and deep disease area expertise, to realize new possibilities in AI-driven drug discovery,” she said.

Image: Isomorphic Labs