Claude Science Autonomously Designs Scorpion Toxin Protein Binders
Anthropic's Claude Science autonomously generated nearly 320,000 protein binder designs targeting scorpion neurotoxins during a 24-hour unattended run. Lab testing confirmed 17 successful binders from the shortlisted candidates, with two showing extremely tight binding to the Cn2 toxin. This achievement marks a significant breakthrough in large language models' ability to design complex biological molecules from scratch, providing strong evidence for AI-assisted drug discovery.
Full Report
Anthropic's Claude largely autonomously designed protein binders targeting scorpion neurotoxins, which kill more than 3,250 people every year. These small protein molecules are shaped to latch onto a specific target molecule. Claude generated 318,484 designs for such protein binders targeting five toxins from four scorpion species. After an initial prompt, Claude worked on its own for 24 hours. The research team then steered the work with additional instructions.
Claude picked 180 candidates, and the team added 31 more. Of the 211 designs tested in the lab, 17 met the criteria for a binder. The two strongest bound extremely tightly to Cn2, a toxin from the Mexican scorpion. Lab tests suggest they dock at or near the same site as an antibody known to neutralize Cn2 in mice. Whether they actually work as an antivenom hasn't been tested yet.
External experts Susana Vázquez Torres and Timothy P. Jenkins called the work a promising first step. Two independent labs confirmed the results. Claude worked inside Claude Science, Anthropic's AI environment for researchers. In an earlier study, Claude had already designed mini-binders against 14 of 15 target proteins, with hit rates of 22 to 35 percent compared to the typical 10 to 15 percent.