Chinese Scientists Develop ContactSeek AI Framework for Precise DNA Base Editing

Researchers from Peking University and East China Normal University have developed ContactSeek, an AI framework that utilizes AlphaFold3's contact probability to design highly specific DNA base editors. Published in Nature, this framework helps identify key amino acid residues to minimize off-target editing effects. Reducing off-target mutations while maintaining high editing activity is a primary challenge in gene editing therapeutics. ContactSeek establishes a new paradigm by combining AI-driven structural modeling with high-throughput sequencing to engineer safer and more precise gene-editing tools. Although AlphaFold3's 3D structural predictions may not directly show conformational changes from base mismatches, its contact probability (CP) output is highly sensitive to these variations. By analyzing CP differences between on-target and off-target complexes, ContactSeek successfully identified novel Cas9 variants with significantly reduced off-target activity.