Prediction of off-target specificity and cell-specific fitness of CRISPR-Cas System using attention boosted deep learning and network-based gene feature
Fig 3
Performances comparison of off-target specificity prediction models, including AUC-ROC and PR-AUC scores of AttnToMismatch_CNN, DeepCpf1, DeepCrispr, Random Forest, and Gradient Boosted trees models in 2 different scenarios.
A)-B) Crispr-Cas12a (Crispr-Cpf1). DeepCpf1 PR-AUC score is not provided in the previous study [33]. C)-F) Crispr-Cas9. C) and D) are the performances with the 5-fold cross-validation method. E) and F) are the performances by leaving three sgRNAs out as test dataset (leave-sgRNAs-out). In B), the PR-AUC of DeepCpf1 was not reported.