Claude AI designs proteins targeting scorpion venom in 64 hours, 17 pass lab tests

Claude AI generates protein candidates at unprecedented speed

Anthropic's Claude AI model designed hundreds of thousands of candidate proteins in 64 hours to target scorpion venom toxins, marking a significant demonstration of AI-driven drug discovery capabilities1

1

2

2

. The experiment, conducted with input from investment research firm Citrini Research and validated through laboratory testing by Adaptyv Bio and Twist Bioscience, focused on designing proteins known as minibinders that could attach to five toxins associated with scorpions in Mexico, Brazil, North Africa and the Middle East2

2

. Citrini Research budgeted $200,000 for the three-month project and worked with an internal Anthropic research model more advanced than the publicly released version1

1

.

Designing proteins to block toxins shows early promise

In August, Citrini Research gave Claude AI a roughly 35-page prompt, which spawned more than 300 sub-agents to design the proteins1

1

. The system used structural information from the Protein Data Bank and AlphaFold's database to identify toxin regions involved in interactions with sodium channels, which help nerve cells transmit electrical signals2

2

. The initial campaign produced 180 designs, with four follow-up prompts growing the pool to 211 candidates sent for synthesis and testing1

1

. Of those 211 designs, 17 bound to their targets across four of the five toxins1

1

. Two designs, Cn2-B1 and Cn2-B2, bound with affinities in the low-nanomolar range, though an attempt to design one molecule that binds all five toxins failed1

1

.

Drug discovery applications remain unproven but encouraging

While the results demonstrate protein design capabilities, Anthropic cautioned that no experiment has tested whether the proteins stop the toxins from working1

1

. Binding to a toxin represents an important early milestone, but does not establish that a candidate can prevent the toxin from causing harm2

2

. External reviewer Timothy P. Jenkins noted that "designing credible binders against five toxins in parallel, in a matter of days, is a great showcase of the AI-driven paradigm shift we are seeing in therapeutic discovery"2

2

. Citrini Research emphasized that "a great deal of science remains before anyone could call the result a treatment"1

1

. Anthropic is releasing the sequences and experimental measurements for all 211 designs tested, allowing other researchers to build on the findings2

2

.

Life sciences stocks positioned to benefit from AI virtual biologists

Citrini Research is backing life sciences stocks following the experiment, with Charles River Laboratories, Revvity and Tecan Group among its top picks1

1

. Citrini analysts argued that growing use of AI virtual biologists should lift demand for "any company that sends them their first real-world invoice," calling Revvity "the compute of biotech"1

1

. The firm said it began looking outside AI infrastructure in early June and settled on life sciences and healthcare, describing the sector as "a secular story at cyclical lows" after being "a graveyard for returns" over the past five years1

1

. Prompting took place at Anthropic's offices because Citrini's request to bypass a biosecurity classifier required security measures it did not have1

1

.

Scorpion venom research addresses global health challenge

Anthropic focused on scorpion envenomation, which causes at least 1.2 million stings and more than 3,250 deaths annually2

2

. Children in regions with limited access to medical care face particular risks, while the high cost of developing new treatments and limited financial incentives create barriers to addressing the problem2

2

. Traditional antivenom production typically involves collecting venom, immunizing animals and purifying the resulting antibodies, a process Anthropic described as resource-intensive and slow2

2

. Computationally designed proteins could eventually be refined for greater stability and potency, potentially opening another route to developing treatments2

2

.

Claude Science workbench expands pharmaceutical partnerships

Anthropic unveiled Claude Science in July, a workbench that connects to more than 60 scientific databases and uses a primary agent that can spawn specialized sub-agents1

1

. It runs on existing Claude models, including Opus 4.8, and Anthropic also started an internal drug discovery program1

1

. Novo Nordisk and Anthropic announced a collaboration in September on drug discovery problems identified by Novo's scientists, with Novo testing Claude Science for specific research and development workflows1

1

. Watch for additional pharmaceutical partnerships as companies seek to accelerate drug discovery timelines and reduce development costs through AI-powered protein design capabilities.