Engineering the X-Editor for Therapeutic-Grade In Vivo Editing with DeepXE, a Predictive Model for Guide Potency

  • We have engineered the compact CRISPR effector CasX (XE) to achieve >100-fold higher activity than wild-type CasX, establishing a potent platform for therapeutic in vivo application
  • The XE platform demonstrates clinical readiness with saturating editing in non-human primate liver models and no detectable off-target activity above background at supersaturating doses in primary human hepatocytes
  • We introduce “DeepXE,” an AI model trained on a quantitative dataset of >40,000 gRNAs; this predictive engine accurately identifies high-efficiency therapeutic guides, dramatically reducing the cycle time for candidate selection
  • We advanced targetability of XE by building models such as DeepXE into our comprehensive engineering pipelines