Machine-Learning-Inferred and Energy-Landscape-Guided Analyses Reveal Kinetic Determinants of CRISPR/Cas9 Gene Editing

The CRISPR-Cas9 nuclease system is currently the most powerful tool in genome engineering. Existing data-driven learning and thermal equilibrium-based modeling cannot capture the kinetic mechanism of the gene cleavage process. This study integrates large-scale genome editing data with a random forest approach—a supervised ensemble learning algorithm—and free energy landscape analysis to uncover the kinetic pathways and pivotal steps that govern cleavage efficiency. Major findings highlight a direct correlation between CRISPR cleavage efficiency and R-loop unfolding rate, establishing the R-loop unfolding rate as the critical factor in gene editing performance. The kinetic model significantly improves over other methods in predicting both on- and off-target gene editing efficiencies. Furthermore, energy landscape analysis reveals a novel three-state mechanism and identifies key nucleotides essential for gene editing. These insights lay the biophysical groundwork for new strategies to enhance CRISPR-Cas9 accuracy and efficiency, offering new avenues for single guide RNA design and target site selection.