An improved CNN-LSTM deep model for Classification of guideRNA in CRISPR-Casl2 System

Gene editing with CRISPR-Cas12 method has immense applications in diverse organisms from bacteria, and plants to humans especially in treating genetic diseases. GuideRNA and Cas enzyme are the two components required by the Cas12 gene editing method. However, the CRISPR-Cas12 system varies depending on the guideRNA sequence used. Therefore, it is important to develop improved methods for designing guideRNA sequences needed for gene editing. Here, we utilized a combination of convolutional neural networks and long short-term memory a type of recurrent neural network (bi- LSTM) model to classify guideRNA as either active or inactive, and the addition of an embedding layer for encoding guideRNA sequence improved model performance. Accuracy, AUC and F1-score were used to evaluate CNN-LSTM and achieved outstanding performance. CNN-LSTM will help in the design of guideRNA for genes with CRISPR-Cas12 system.