AI Models Accelerate CRISPR Genome Editing
A new paper in *Nature Reviews Genetics* details how artificial intelligence is advancing CRISPR-based genome editing. The research, co-authored by Xaira Therapeutics SVP Bo Wang, outlines AI's role in guide RNA design, CRISPR enzyme engineering, and the use of virtual cell models to predict editing outcomes for biologics development.
A primary challenge AI addresses in CRISPR is mitigating off-target effects and improving editing efficiency, which are significant hurdles in translating therapies to the clinic. Machine learning models analyze vast genomic datasets to enhance guide RNA (gRNA) design, improving target specificity and reducing unintended mutations—a critical factor for safety and regulatory approval in gene therapy manufacturing. Beyond just optimizing existing components, AI is now being used to create entirely novel gene editors from scratch. Profluent Bio's OpenCRISPR-1, for instance, is an AI-generated enzyme designed by large language models trained on extensive CRISPR datasets, demonstrating comparable efficiency to natural Cas9 but with enhanced specificity. The concept of "virtual cells" involves creating digital twins to simulate cellular responses to genetic modifications *in silico*. This allows for high-throughput screening of potential edits and prediction of Critical Quality Attributes (CQAs) before extensive wet lab experimentation, directly accelerating process development for viral vector production. For biomanufacturing, this AI-driven precision is foundational for developing robust, automated processes. By minimizing variability at the design stage, it simplifies the path to scalable GMP production, enhances data integrity for electronic batch records, and supports the creation of digital twins for entire manufacturing workflows. Bo Wang's work on scGPT, a foundation model for single-cell biology, informs his role at Xaira Therapeutics. Xaira aims to build a comprehensive platform that integrates multimodal AI models across the entire R&D pipeline, from initial target identification to stratifying patients for clinical trials.
