PerturbAI Launches with CRISPR Atlas

By Bio-IT World Staff

March 20, 2026 | PerturbAI unveiled the world’s largest in vivo CRISPR atlas this week as it emerged from stealth. The 8 million-cell, brain-wide dataset represents the largest known causal genomics resource generated directly in living tissue.

Unlike traditional in vitro or single-organ systems, PerturbAI’s in vivo CRISPR (Perturb-seq) platform measures gene function within intact organs, capturing real biological circuitry rather than isolated cell states. The release establishes a new category of biological data: organism-level, circuit-resolved, causal genomics, the company said.

“The public release of this CRISPR atlas marks a foundational change for biology,” said Xin Jin, Ph.D., co-founder of PerturbAI, in a press release. “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.”

“Using large-scale CRISPR screening and single-nucleus RNA sequencing, we’ve built a functional map of the mouse brain's genome,” the company wrote in an un-bylined blog post. “Measuring the effects of nearly 2,000 disease-linked genes in their native environment, we’ve revealed the molecular logic of the neuronal circuits underlying neurodegeneration, psychiatric, and metabolic diseases.”

PerturbAI is using its in vivo CRISPR platform and causal AI models to develop best-in-class therapeutics for 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,” said Grace Zheng, Ph.D., co-founder and CEO of PerturbAI in the same statement.

PerturbAI collaborated with NVIDIA, the Allen Institute for Brain Science, and 10x Genomics to generate and analyze the atlas, which enables systematic evaluation of the brain genome, providing a framework to distinguish causal mechanisms from context-dependent modifiers of disease risk and progression.

“This study demonstrates that massively parallel genetic perturbation profiling in the intact brain can resolve the functional architecture of neuronal gene regulation at single-cell resolution,” authors wrote in a bioRxiv preprint. “Scaling Perturb-seq to nearly 2,000 genes across the mouse brain was enabled by our optimized in vivo delivery, improved gRNA recovery through molecular vector designs, and advances in combinatorial indexing-based single-cell sequencing, building on recent efforts to expand the throughput of perturbation screens and the ‘virtual cell’ efforts.” The team used GPU-accelerated analysis pipelines.

They found that, “the transcriptional consequences of many disease-relevant genes in the brain are remarkably sparse and cell-type-specific – a level of resolution that is largely invisible to bulk tissue profiling,” the authors said in the paper.

“A natural next step is to ask whether perturbing convergent targets can rescue or reverse the phenotypic effects of individual risk-gene mutations, and to extend this approach to additional organs and model systems where the same scalable delivery and sequencing framework can be applied,” they concluded.

Data generated for this study are available through the Hugging Face repository and the UCSC Cell Browser. The analysis pipeline is deposited on GitHub repository.

PerturbAI closed an oversubscribed pre-seed round in Q3 2025, backed by leading early-stage investors with deep expertise across AI, biology, and company-building.