Genome editing with CRISPR-Cas9 technology has enabled unprecedented effectiveness in reverse genetics and gene correction. However, while off-target effects and limited target site selection are problems that have been successfully tackled, only minor steps have been taken in the effort to eliminate variability in sgRNA efficacies. Our latest paper, “Refined sgRNA efficacy prediction improves large- and small-scale CRISPR-Cas9 applications“, features the work of Maurice Labuhn, and addresses this issue using an unique approach involving the large-scale assessment of individual sgRNA efficacies at the single-cell level. Moreover, this study highlights current limitations and endeavors in the topic of sgRNA efficacy prediction.
Based on a self-generated dataset of 430 experimentally quantified sgRNAs, the authors tested the predictive value of 5 recently-established prediction algorithms. Only a mild correlation (R = 0.04 to 0.20) was observed between the predicted and measured sgRNA activities, as well as poor concordance between the different algorithms. Therefore, the authors developed CRISPRater, a linear model-based discrete system derived from 10 predictive sgRNA sequence features including the PAM-distal GC content. CRISPRater was verified on small- and large-scale external datasets, and achieves an improved efficiency in selecting effective sgRNAs (R = 0.40). It is now integrated into the online sgRNA design tool CCTop, providing a versatile combined on- and off-target sgRNA scanning platform.