Isomorphic Labs Reveals IsoDDE: Scientists Call It an 'AlphaFold 4-Scale' Drug AI

Overview

Isomorphic Labs, Google DeepMind's drug discovery spin-off, released a 27-page technical report in February 2026 describing IsoDDE (Isomorphic DeepMind Drug Engine)—a proprietary AI model for predicting molecular interactions in drug development. Researchers in the field have reacted with excitement, calling it a landmark advance on par with unreleased future generations of AlphaFold.

Key Capabilities

• Protein-ligand interaction prediction: Highly accurate forecasting of how drug molecules bind to target proteins, including binding affinity scores
• Antibody structure prediction: Modeling antibody-target interactions for biological therapeutics
• Performance: Outperforms open-source alternatives and physics-based computational methods across key benchmarks

Scientific Reception

Mohammed AlQuraishi, a computational biologist at Columbia University, told Nature: "It is a major advance, on the scale of an AlphaFold4." The comparison to an unreleased future AlphaFold generation reflects the model's outsized impact on a field where drug-binding prediction accuracy directly shapes development timelines. Antibody-based therapies alone generate tens of billions in annual pharmaceutical revenue.

Controversy: Proprietary vs. Open

Unlike the AlphaFold models that DeepMind published openly—enabling global research—IsoDDE is strictly proprietary. The technical report offers minimal methodological detail, triggering debate about scientific transparency versus commercial incentives.

Commercial Partnerships

Isomorphic Labs has already struck drug-development deals potentially worth billions of pounds with pharmaceutical majors Johnson & Johnson, Eli Lilly, and Novartis. The technical report substantiates the computational foundation underpinning these partnerships.

Source: Scientific American | Nature