Thousands of AI Agents Analyze Brain Cell Data in 32 Minutes!

Hi everyone, Shii-chan here! Today's story is about AI speeding up biology research in a big way.

OpenAI News openai.com

What was announced?

OpenAI News describes a collaboration between OpenAI, PerturbAI, and Scripps Research that used thousands of AI agents to find research leads in complex biological data. The agents, powered by GPT-Rosalind, analyzed existing experimental data from millions of mouse brain cells. The analysis took 32 minutes. The team estimates that comparable human analysis would take months to a year.

Why it matters

Experiments can now disrupt genes one by one and record the effect on cells at huge scale. But the data is so large that no human can read the literature and interpret every gene. This work scales that interpretation step with parallel agents.

What changes

• Researchers can get promising candidates from huge datasets in a short time
• A "reasoning atlas" links genes, cellular effects, relevant publications, and possible mechanisms, so you can pick which hypotheses to chase
• The atlas is public on GitHub (jinlabneurogenomics/distributed_agent), so other researchers can use it

Dive Deep

The dataset is a whole-brain CRISPR atlas showing how disrupting individual genes affects different cell types across the mouse brain.

The method is called "distributed biological reasoning":

• Each agent takes one gene disruption
• It analyzes the cellular responses, searches the scientific literature, and proposes biological explanations
• The agents' reports are merged into the reasoning atlas

The agents reproduced findings from human analyses and found additional leads. Their judgments of novelty and scientific value matched those of blinded human experts.

Two examples were highlighted:

• Tsc1 vs. Tsc2: Both regulate mTOR, a central controller of cell growth and metabolism, and disrupting either had similar effects in several neuron groups. In the cerebellum, though, Tsc2 disruption changed dozens of genes involved in neuronal communication while Tsc1 disruption caused little change. That's relevant to tuberous sclerosis, where cerebellar abnormalities are linked to TSC2 mutations, and it points to specific questions about existing drugs and new approaches
• DCTN1 and cholesterol: DCTN1 mutations cause Perry syndrome, a neurodegenerative disease with Parkinson-like symptoms. Disrupting Dctn1 increased expression of cholesterol-synthesis genes. The agents tied this to research suggesting transport problems can make cells overproduce cholesterol, and hypothesized that DCTN1 disruption alters cholesterol distribution in neurons and contributes to neurodegeneration. Human experts judged it novel and worth testing

There's also a comparison with a single agent:

• Given the full dataset with the same model and resources, a single agent cited only 11 references at the largest workload
• Distributed agents cited more than 700 references
• In blinded comparisons, human experts preferred distributed-agent reports 75% of the time

Wrap-up

• Thousands of agents analyzed data from millions of mouse brain cells in 32 minutes
• One agent per gene disruption, merged into a reasoning atlas
• Distributed agents cited more literature (700+ vs. 11) and won 75% of expert comparisons
• The atlas is public on GitHub

If you care about genomics, neuroscience, or how to scale agents, this one's for you!