PerturbAI Launches With Massive In Vivo CRISPR Brain Atlas, Aiming to Remap Drug Discovery

SAN FRANCISCO, CA – March 17, 2026 – A new player in the biotech arena, PerturbAI, emerged from stealth today, unveiling a monumental dataset that could fundamentally alter how new medicines are discovered. The AI-native therapeutics company announced the public release of the world's largest in vivo CRISPR atlas, a massive 8-million-cell map detailing gene function across the entire brain of a living organism.

This brain-wide dataset, generated in a high-profile collaboration with AI giant NVIDIA, the Allen Institute for Brain Science, and genomics leader 10x Genomics, represents the largest known causal genomics resource ever created directly within living tissue. By moving beyond traditional laboratory dishes, PerturbAI aims to capture the complex, interconnected reality of biology, promising to make therapeutic development faster, cheaper, and more likely to succeed.

A New Paradigm for Biological Discovery

For decades, much of biological research has relied on in vitro models—studying cells in isolation in a petri dish. While foundational, this approach fails to capture the intricate web of interactions that occur within a complete, living organism. PerturbAI is pioneering a different path with its in vivo CRISPR platform, a technology that uses the gene-editing tool CRISPR to systematically switch genes off and on directly inside a living system.

By combining this with advanced single-cell sequencing, a technique known as Perturb-seq, the company can observe the precise effects of thousands of genetic changes across hundreds of different cell types in their natural environment. The result is what the company calls a new category of data: "organism-level, circuit-resolved, causal genomics." This allows scientists to move from correlation to causation, directly linking a specific gene to its function within the complex circuitry of an organ like the brain.

"The public release of this CRISPR atlas marks a foundational change for biology," said Xin Jin, Ph.D., co-founder of PerturbAI, in a statement. "For the first time, we can measure gene function directly inside intact biological systems at a massive scale. This enables systems-level understanding of disease and allows us to model and simulate therapeutic interventions before committing to expensive downstream drug development."

The initial atlas focuses on the mouse brain, revealing that the same gene can have vastly different effects depending on the cell type, brain region, and biological context—a level of nuance often lost in simpler models.

The Race for Foundational Genomic Atlases

PerturbAI's launch places it at the forefront of a critical trend in biotechnology: the creation of massive, foundational datasets to fuel AI-driven discovery. While its 8-million-cell in vivo atlas appears to be the largest of its kind, it enters a competitive landscape of ambitious data-generation projects.

Earlier this year, genomics powerhouse Illumina announced its own effort to build the "world's largest genome-wide genetic perturbation dataset" in partnership with pharmaceutical giants like AstraZeneca and Merck. However, that project focuses on perturbing genes across 250 different human cell types grown in labs, with a goal of reaching one billion cells. The key distinction lies in the environment: PerturbAI’s data comes from a living organism, while Illumina’s comes from cell cultures.

Both approaches are seen as vital. In vitro atlases provide a massive, scalable reference using human cells, ideal for high-throughput screening. In contrast, in vivo atlases like PerturbAI's, while currently on a smaller scale, offer unparalleled insight into systemic effects, immune responses, and organ-level biology that can never be fully replicated in a dish. This provides crucial context for understanding how a potential drug might behave in a patient.

A Trifecta of Technology, Neuroscience, and AI

Generating and analyzing a dataset of this magnitude was not a solo effort. PerturbAI’s launch was bolstered by a strategic collaboration with three industry titans. 10x Genomics, whose technology is a cornerstone of the Perturb-seq method, provided the essential tools for single-cell analysis at scale. The Allen Institute for Brain Science, a world leader in mapping the brain's cellular and genetic architecture, brought invaluable neuroscientific expertise to the project.

Completing the trifecta is NVIDIA, whose computational power was essential for processing the terabytes of data produced. The atlas is now accessible through the NVIDIA AI Blueprint for Single-Cell Analysis, highlighting its role as a key training resource for next-generation AI. By training AI on this rich causal data, researchers can build powerful biological foundation models capable of predicting gene function and simulating the effects of new drugs with unprecedented accuracy.

This collaborative approach underscores a growing reality in modern science: breakthroughs of this scale require a deep integration of biology, hardware, and artificial intelligence.

From Atlas to Therapeutics: The Road Ahead

While the public release of the brain atlas is a major scientific contribution, PerturbAI's ultimate goal is commercial. The company intends to leverage its platform to develop its own pipeline of best-in-class therapeutics, initially targeting complex metabolic and chronic diseases.

"Given our team's deep expertise spanning CRISPR, genomics, and machine learning, we are uniquely qualified to optimize the application of vivo CRISPR and new AI models to novel therapeutic discovery, making it more scalable, translatable and cost-efficient," stated Grace Zheng, Ph.D., co-founder and CEO of PerturbAI.

The company’s strategy appears to be resonating with investors. PerturbAI confirmed it closed an oversubscribed pre-seed funding round in the third quarter of 2025 from what it described as "leading early-stage investors," though it did not disclose the amount or the backers.

Despite the promise, the road from an atlas to an approved drug is long and fraught with challenges. The in vivo application of CRISPR technology itself faces significant hurdles, including ensuring the safe and efficient delivery of gene-editing components to target tissues and minimizing the risk of off-target effects. However, by creating a high-fidelity map of causal biology, PerturbAI is betting it can significantly de-risk the process, identifying the most promising therapeutic targets before a single dollar is spent on clinical trials. The company's public debut signals a bold step toward a future where drug discovery is less a matter of chance and more a function of precise, data-driven engineering.