AI-driven contact modelling sharpens genome editing accuracy with AlphaFold3
ContactSeek, an AI framework, leverages AlphaFold3-predicted contact probabilities to map on- and off-target interactions of genome-editing complexes (Cas9–TadA8e and Cas12a-based CBEs). By correlating contact changes with sequencing-based off-target signals, it identifies consensus contact regions and specificity-determining residues, enabling targeted mutations that greatly improve editing specificity while preserving on-target activity. The approach, which generalizes to multiple editors, establishes an AF3-driven paradigm for enhancing precision in genome editing by integrating structural and functional dimensions.