Systematic decomposition of sequence determinants governing CRISPR/Cas9 specificity

The specificity of CRISPR/Cas9 genome editing is largely determined by the sequences of guide RNA (gRNA) and the targeted DNA, yet the sequence-dependent rules underlying off-target effects are not fully understood. Here we systematically investigated the sequence determinants governing CRISPR/Cas9 specificity by measuring the off-on ratios of 1,902 gRNAs on 13,314 target sequences using an improved synthetic system with dual-target design. Our study revealed a comprehensive set of rules including 3 factors in CRISPR/Cas9 off-targeting: 1) the nucleotide context and position of a single mismatch; 2) an "epistasis-like" combinatorial effect of multiple mismatches; and 3) a guide-intrinsic mismatch tolerance (GMT) independent of the mismatch context. Notably, the combinatorial effect and GMT are associated with the free-energy landscape in R-loop formation and are explainable by a multi-state kinetic model. Based on these rules, we developed a model-based off-target prediction tool (MOFF), which showed superior performance compared to the existing methods.