An Artificial Intelligence Approach for Gene Editing Off-Target Quantification: Convolutional Self-attention Neural Network Designs and Considerations

In the CRISPR-based gene-editing system, an important issue is the off-target cleavage which could alter the functions of unintended genes and induce toxicity. Numerous biological techniques have been proposed to detect the off-target effects. However, those laboratory-based techniques are expensive and time-consuming for guide RNA selection. Therefore, we introduce a computational method based on convolutional neural network and attention module to predict the CRISPR off-target activity. With two validation experiments, we demonstrate that our proposed model has improved predictive performance over the state-of-the-art deep-learning-based off-target prediction models in terms of Receiver Operating Characteristics and Precision-Recall analyses. For scientific reproducibility, we have made the source code available at the GitHub repository (https://github.com/JasonLinjc/CRISPRattention).

Automated summary

The issue of off-target cleavage in the CRISPR gene-editing system has been a concern. This study introduces a computational method using convolutional neural network and attention module to predict off-target activity in CRISPR. Validation experiments demonstrate that the proposed model outperforms existing deep-learning-based off-target prediction models in terms of predictive performance.