‘An AlphaFold 4’ – scientists marvel at DeepMind drug spin-off’s exclusive new AI

2026-02-19 · Source: Machine learning : nature.com subject feeds · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, AI for Drug Discovery · Depth: Advanced, quick

Summary

On February 19, 2026, Isomorphic Labs, a biopharmaceuticals spin-off of Google DeepMind, announced a new proprietary artificial intelligence model called IsoDDE, detailed in a 27-page technical report released on February 10. This model, described by some scientists as an "AlphaFold4"-scale advance, significantly improves predictions of how proteins interact with potential therapeutic molecules and antibody structures. Unlike previous AlphaFold systems, IsoDDE's underlying methodology remains undisclosed, leaving open-source researchers unable to replicate its performance. The model reportedly outperforms both open-source alternatives like Boltz-2 and traditional physics-based methods in determining drug-protein binding affinity, a critical factor in drug development, and excels at predicting interactions for molecules vastly different from its training data.

Key takeaway

For entrepreneurs and investors evaluating drug discovery platforms, Isomorphic Labs' IsoDDE represents a significant, albeit proprietary, leap in AI capabilities. You should consider its reported superior performance in drug-protein interaction and binding affinity prediction when assessing potential partnerships or competitive landscapes. However, the lack of transparency means you cannot replicate or build upon its core methods, which could influence long-term strategic decisions regarding open-source versus proprietary tool adoption.

Key insights

Isomorphic Labs' proprietary IsoDDE model significantly advances AI drug discovery, outperforming open-source alternatives in binding affinity prediction.

Principles

• Proprietary AI can outpace open-source efforts.
• Predicting drug-protein binding affinity is crucial.

In practice

• Evaluate IsoDDE for drug-protein interaction predictions.
• Compare IsoDDE against Boltz-2 for binding affinity.

Topics

• Isomorphic Labs
• Drug Discovery AI
• Protein-Drug Interactions
• Binding Affinity Prediction
• Proprietary AI Models

Best for: Investor, Entrepreneur, AI Researcher, AI Scientist, Research Scientist

Related on AIssential

• How a Google DeepMind Spinoff Hunts Hidden Drug Targets — IsoDDE enables discovery of novel, cryptic protein binding pockets, expanding druggable targets beyond AlphaFold's capabilities.
• 😺 🎙️ Watch: How Isomorphic Labs works to drug "undruggable" diseases — AI, particularly AlphaFold-derived technology, is dramatically accelerating drug discovery and tackling previously "undruggable" targets.
• DeepMind spinout Isomorphic Labs raises $2.1bn — AI-driven drug discovery benefits from specialized teams and distinct corporate structures.
• Why We're Not Close to AI Drugs Yet | Isomorphic Labs — Drug design comprises multiple distinct ML problems, not a single one, despite AlphaFold 2's protein structure prediction success.
• AWS and Hopkins Engineering announce groundbreaking database for AI/ML antibody design — A new diverse, experimentally validated dataset enables rigorous benchmarking of AI models for antibody developability.
• AI is reengineering drug discovery by speeding up testing and scanning petabytes of data for connections between diseases — AI significantly enhances drug discovery by analyzing vast data to predict targets, efficacy, and disease interrelationships, improving success rates.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.