CNN-SVR

CNN-SVR predicts guide RNA (gRNA) on-target cleavage efficacy for the CRISPR/Cas9 genome editing system using a hybrid convolutional neural network and support vector regression approach.

Key Features:

• Improved CNN Architecture: A refined convolutional neural network architecture extracts intricate features from input data to capture complex sequence patterns.
• Sequence and Epigenetic Feature Extraction: The CNN processes gRNA sequences and associated epigenetic data to learn deeper representations and feature interactions.
• Support Vector Regression (SVR): An SVR model performs regression on features extracted by the CNN to predict gRNA cleavage efficiency.
• Performance and Robustness: Experimental evaluation on commonly utilized datasets indicates improved prediction accuracy, generalization capability, and robustness compared to prior methods.

Scientific Applications:

• gRNA on-target efficacy prediction: Predicts on-target cleavage efficiency of gRNAs for CRISPR/Cas9 genome editing experiments.
• gRNA design optimization: Supports selection and prioritization of candidate gRNAs based on predicted cleavage efficacy to inform gene-editing experiment design.

Methodology:

The method uses a convolutional neural network to extract features from gRNA sequences and epigenetic data, feeds those features into a support vector regression model for efficacy prediction, and evaluates performance on commonly utilized datasets.