sgRNACNN

sgRNACNN predicts sgRNA on-target activity for CRISPR/Cas9 in Glycine max (soybean), Zea mays (corn), Sorghum bicolor, and Triticum aestivum (wheat) using convolutional neural networks to support efficient gene editing in these crops.

Key Features:

• Ensemble Convolutional Neural Networks: Employs an ensemble of convolutional neural networks combined via a voting algorithm to improve prediction robustness and accuracy.
• Sequence Encoding: Uses one-hot encoding and k-mers to convert sgRNA sequence data into machine-learning input features.
• Reported Prediction Accuracy: Demonstrates prediction accuracies of 82.43% for Glycine max, 80.33% for Zea mays, 78.25% for Sorghum bicolor, and 87.49% for Triticum aestivum.

Scientific Applications:

• sgRNA selection for CRISPR/Cas9 experiments: Guides selection of high on-target activity sgRNAs for genome editing in the four specified crop species.
• Crop trait improvement: Supports design of edits to modify traits such as yield, disease resistance, and stress tolerance in agronomic species.
• Breeding and agricultural biotechnology: Provides predictive data to inform crop improvement programs and applied biotechnology research.

Methodology:

Sequence data are encoded with one-hot encoding and k-mers and input to an ensemble of convolutional neural networks whose outputs are combined via a voting algorithm for on-target activity prediction.