- Published on
Claude Science tackles deadly scorpion stings by designing protein binders from scratch
- Authors
- Name
- DeepHorus
Anthropic’s Claude Science has autonomously designed protein binders aimed at scorpion neurotoxins, which kill more than 3,250 people each year. The system generated 318,484 candidate binders against five toxins from four scorpion species. After an initial prompt, Claude worked alone for 24 hours before researchers steered it with further instructions.
Protein binders are small molecules shaped to latch onto specific targets. The project sought binders that could attach to dangerous scorpion toxins and potentially interrupt their effects. This matters because antivenom access and effectiveness remain limited in many affected regions.
Claude selected 180 candidates, and the research team added 31 more, producing 211 designs for laboratory testing. Of those, 17 met the criteria for a binder. The two strongest candidates bound extremely tightly to Cn2, a toxin from the Mexican scorpion.
Lab tests indicated the top binders dock at or near the same site as an antibody known to neutralize Cn2 in mice. Whether the designs actually work as an antivenom has not yet been tested. Two independent labs confirmed the results, according to the report.
External experts Susana Vázquez Torres and Timothy P. Jenkins described the work as a promising first step. Their assessment underscores both the novelty and the early stage of the research. The findings still require validation in relevant biological models.
Claude operated within Claude Science, Anthropic’s AI environment for researchers. In an earlier study, Claude designed mini-binders against 14 of 15 target proteins, with hit rates of 22 to 35 percent. That compared with typical rates of 10 to 15 percent.
The scorpion-toxin project suggests AI can accelerate early-stage protein design, but it does not yet prove a usable therapy. Future work must test efficacy, safety, and delivery in animals and eventually humans. For now, the results mark a computational advance with potential public-health relevance.
