Zero-shot design of drug-binding proteins via neural iterative selection−expansion

Researchers designed drug-binding proteins from scratch using a novel neural iterative selection-expansion algorithm, as published in Nature on June 24, 2026. This method employs two neural networks in an iterative optimization process to achieve high accuracy, affinity, and success rates in protein design. The developed technique allows for the creation of small-molecule binding proteins without relying on existing templates, a significant advancement in de novo protein engineering. Initial experiments demonstrated the algorithm's capability to generate functional proteins with desired binding properties. The potential applications of this technology are substantial, particularly in areas such as targeted drug delivery, where precisely engineered proteins can bind to specific molecules within the body, and in environmental remediation for sequestering harmful substances. The iterative nature of the algorithm allows for continuous refinement of protein structures, leading to optimized binding characteristics. This breakthrough offers a powerful new tool for synthetic biology and pharmaceutical development, potentially accelerating the discovery of novel therapeutics and biotechnological solutions.