Fig. 3.

Evaluation of CGD using independent test data. Comparison of existing machine learning algorithms, including deep learning algorithms, and our CGD models on independent test data using the KS test. The numbers of efficient (Eff) and inefficient (Ineff) gRNAs are indicated on the x-axis for machine learning (CGD9, CGDi/a, CGD12a, CINDEL 1.0, SSC, and CRISPRko v1/2) or deep learning (DeepCas9/Cpf1) algorithms. a–d Box plots representing the classification efficiency of different algorithms with P values as an indicator of competence for the respective datasets for CRISPRi (a), CRISPRa (b), Cas9 (c), and Cas12a (d).