OpenAI, PerturbAI, Scripps use GPT-Rosalind agents to mine brain data in 32 minutes

The work addresses a practical bottleneck in modern biology: experiments can now alter thousands of genes and measure effects across millions of cells, generating more data than any team can thoroughly review. By dividing the analysis among thousands of GPT-Rosalind agents, each investigating one gene disruption against the scientific literature, the team produced a reasoning atlas that links genes, cellular effects, publications and possible mechanisms.

The comparison with a single agent is telling. When given the entire dataset, the single agent searched less of the literature, citing only 11 references at the largest workload versus more than 700 for the distributed agents, and human experts preferred the distributed reports in blinded comparisons. That suggests the benefit comes from parallel, focused reasoning rather than raw model capability alone.

The biological leads themselves point to specific follow-up experiments. The Tsc1/Tsc2 difference in the cerebellum may be relevant to tuberous sclerosis, where TSC2 mutations are strongly linked to cerebellar abnormalities, and the DCTN1 finding raises the question of whether disrupted cholesterol distribution contributes to neurodegeneration. Human experts considered the DCTN1 hypothesis novel and worth pursuing, but the body is clear that these remain hypotheses for lab testing rather than confirmed treatments.