[Prediction of CRISPR/Cas9 off-target activity using multi-scale convolutional neural network].

• DOI: 10.13345/j.cjb.230382
• Publication Date: 2024-03-25
• Authors: Huanzeng Xie; Lingze Huang; Ye Luo; Guishan Zhang
• Keywords: Guide RNA; ENCODE

Clustered regularly interspaced short palindromic repeat/CRISPR-associated protein 9 (CRISPR/Cas9) is a new generation of gene editing technology, which relies on single guide RNA to identify specific sites and Cas9 nuclease edit location in the genome. However, off-target effect this technology hampers its development. In recent years, several deep learning models have been developed for predicting CRISPR/Cas9 activity, contributes more efficient safe therapy. prediction accuracy remains be improved. paper, we proposed multi-scale convolutional neural network-based method, designated as CnnCRISPR, prediction. First, used one-hot encoding method encode sgRNA-DNA sequence pair, followed by bitwise or operation two binary matrices. Second, encoded was fed into Inception-based network training evaluating. Third, well-trained model applied evaluate situation pair. Experiments public datasets showed CnnCRISPR outperforms existing learning-based methods, provides an effective feasible addressing problems.