Off-target sequence variations driven by the intrinsic properties of the Cas–sgRNA–DNA complex in genome editing

Genome-editing technologies hold significant potential across various biotechnological fields, yet concerns about possible risks, including off-target mutations, remain. To ensure safe and effective application, these unintended mutations must be rigorously examined and minimized. Computational approaches are anticipated to streamline the detection of off-target mutations; however, the performance of current prediction tools is limited, likely owing to insufficient knowledge of off-target mutation characteristics. In this study, we collected experimentally validated off-target mutation data and conducted a large-scale analysis of 177 nonredundant datasets obtained from six studies. We developed a method to assess the statistical significance of sequence pattern similarity and diversity between off-target sites. This method is based on a comparison of ordered relative entropy values for aligned target sequences, and it was compared with two other methods on the basis of Euclidean distance and the Pearson correlation coefficient. The three methods demonstrated clear correlations, indicating their validity. These methods were applied to 238 dataset pairs for the same target site, and it was revealed that off-target sequence patterns were quite similar across different experimental conditions, such as varying cell lines and independent experiments, suggesting that the intrinsic properties of the Cas–sgRNA–DNA complex play a key role in determining cleavage sites. However, newly engineered enzymes and those from different bacterial sources occasionally display unique off-target patterns, indicating the need for comprehensive evaluation of each new enzyme to develop reliable prediction tools. The insights gained from this study are expected to contribute to a better understanding of off-target mutation characteristics and support the development of more accurate computational prediction methods.