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

Nature

by Ewen Callaway

February 20, 2026

AI-Generated Deep Dive Summary

Scientists are impressed by Isomorphic Lab’s new AI model, IsoDDE, which represents a significant leap in drug-discovery capabilities. The proprietary tool, revealed in a technical report, excels at predicting how proteins interact with potential therapeutic molecules and antibody structures, outperforming both open-source models like Boltz-2 and traditional physics-based methods. Unlike previous AlphaFold iterations, which were made accessible to researchers, IsoDDE’s details remain closely guarded, leaving the scientific community eager for insights. The development of proprietary AI tools like IsoDDE reflects a growing trend in biotech, where companies are prioritizing confidentiality over collaboration. While Isomorphic Labs highlights its model’s ability to predict interactions with molecules far removed from its training data as a key innovation, this lack of transparency contrasts with open-source projects like AlphaFold, which have fostered widespread research advancements. The implications for drug discovery are profound. IsoDDE’s reported success in predicting binding affinities and antibody-target interactions could accelerate the development of new therapies. However, its proprietary nature raises concerns among scientists like Mohammed AlQuraishi, who emphasizes the importance of open-source tools in fostering innovation and democratizing access to cutting-edge technology. This shift toward closed-source AI models underscores a broader debate in scientific research: whether proprietary solutions can truly advance knowledge as effectively as collaborative efforts. While IsoDDE’s capabilities are undeniably impressive, its lack of transparency leaves many researchers wondering how they might replicate or build upon such breakthroughs. Ultimately, the race to develop advanced AI tools for drug discovery is reshaping the field. While proprietary models like IsoDDE offer potential benefits in efficiency and speed, the scientific community will need to weigh these gains against the limitations of limited access and collaboration. The future of AI-driven drug discovery may hinge on finding a balance between innovation and

Verticals

scienceresearch

Originally published on Nature on 2/20/2026