Using thousands of AI agents to uncover new leads in biological data

OpenAI, PerturbAI, and Scripps Research deployed thousands of AI agents powered by GPT‑Rosalind to analyze data from millions of mouse brain cells in 32 minutes, a task estimated to take human experts months to a year. The agents connected gene disruptions with scientific literature, uncovering unexpected biological connections that may aid disease research and treatment development.

The team used a dataset called a whole-brain CRISPR atlas, where each AI agent examined a gene disruption, analyzed cellular responses, and searched relevant publications to propose explanations. Their combined reports formed a reasoning atlas linking genes, cellular effects, publications, and potential biological mechanisms for further investigation.

The agents identified unexpected differences in how two disease-linked genes, Tsc1 and Tsc2, affect brain cells, particularly in the cerebellum, which may be relevant to human conditions like tuberous sclerosis. They also found a potential connection between cholesterol regulation and neurodegeneration, highlighting new research directions.

The research showed that distributed AI agents cited over 700 references compared to 11 by a single agent, and human experts preferred distributed-agent reports 75% of the time. The reasoning atlas is publicly available, enabling scientists to explore findings and prioritize experiments based on AI-generated leads.