Profluent Announces Publication of Generative AI Research in Nature with New Results for OpenCRISPR-1

July 30, 2025 at 11:00 AM EDT

New results show impressive activity, specificity, and immunogenicity for Profluent’s language model-generated CRISPR-Cas proteins, which outperform structure-based and baseline methods
Following broad adoption of OpenCRISPR-1, Profluent to open-source the CRISPR-Cas Atlas – the most extensive dataset of CRISPR systems to date – to further democratize access to next-generation gene editing
Profluent has scaled its gene editing capabilities across modalities to become the preferred partner for bespoke gene editor needs

Profluent, the AI-first protein design company, today announced a publication in Nature describing the design of OpenCRISPR-1, the world’s first AI-created genome editor to successfully edit the human genome. These findings validate the use of Profluent’s large language models (LLMs) to generate custom gene editors that match or exceed the performance of naturally occurring CRISPR systems, with the potential to expand gene editing capabilities and accelerate therapeutic development.

Despite the diversity of CRISPR-Cas proteins that exist in nature, not all are ideally suited for specific therapeutic applications, and traditional protein engineering approaches introduce tradeoffs that limit utility. The publication, titled “Design of highly functional genome editors by modelling CRISPR–Cas sequences,” explores Profluent’s approach to AI-enabled protein design as a compelling alternative to directed evolution and structure-guided mutagenesis for engineering novel gene editing systems at scale.

“As our Nature publication demonstrates with the highest scientific rigor, the novelty and diversity of the gene editors Profluent can create with our LLMs transcend that of proteins made by nature or traditional engineering,” said Ali Madani, Ph.D., Profluent co-founder and Chief Executive Officer. “By opening up this new frontier in AI-powered drug discovery, we are shifting biology from a field constrained by evolution to one of abundant possibilities, where we can generate new proteins on demand to solve society’s most pressing challenges.”

The paper expands on a previously published preprint with several new findings that reinforce the effectiveness of AI for protein design:

AI generates sequences with unprecedented novelty and diversity: LLMs trained on large-scale biological sequences produce diverse CRISPR-Cas sequences spanning a broad set of families, while being hundreds of mutations away from known natural proteins and therefore impossible to create manually.
Viability of LLMs for CRISPR-Cas design: Profluent’s platform performed exceptionally well in designing functional Cas proteins compared to manual or other machine learning approaches to protein engineering tested.
Genome-wide off-target analysis of OpenCRISPR-1: OpenCRISPR-1 showed significantly reduced off-target cleavage events using an unbiased genome-wide specificity assay.
Lower potential immunogenicity with OpenCRISPR-1: iELISA quantification showed lower immunogenicity for all generated Cas proteins compared to SpCas9, suggesting that new-to-nature proteins designed with machine learning have the potential to be less immunogenic as compared to pathogen-derived genome editors, like SpCas9.

“Since releasing OpenCRISPR-1 in 2024, tens of thousands of academic and industry researchers have accessed the open-source sequence, signaling AI’s broad potential to unlock solutions across therapeutics, biomanufacturing, agriculture, climate, and beyond,” said Hilary Eaton, Ph.D., Profluent Chief Business Officer. “As we deepen investment in our platform capabilities to meet the bespoke needs of our partners, we’re building Profluent into a one-stop-shop for customized gene editors of any modality, with attractive economics and flexible partnering models.”

OpenCRISPR-1 is publicly available and free to license for ethical research and commercial use. The open-source package can be accessed via GitHub and AddGene. Profluent has also decided to open-source the CRISPR-Cas Atlas, the most extensive dataset of CRISPR systems curated to date, that was used to steer Profluent’s models. This move aims to further expand access to Profluent’s next-generation gene editing capabilities.

About Profluent

Profluent is an AI-first company pushing the frontier of de novo protein design to author new biology. Grounded in nature with AI as an interpreter, Profluent’s powerful foundation model platform unlocks solutions that transform medicine, agriculture, and beyond. Founded in 2022 and headquartered in Emeryville, CA, Profluent is backed by leading investors including Spark Capital, Insight Partners, Air Street Capital, AIX Ventures, and Convergent Ventures. To learn more, visit profluent.bio.