CNN-Pred

CNN-Pred classifies proteins as single-stranded DNA-binding proteins (SSBs) or double-stranded DNA-binding proteins (DSBs) using evolutionary features and a 1D-convolutional neural network to improve prediction accuracy.

Key Features:

• Target classification: Classifies proteins into SSBs and DSBs.
• Evolutionary-based features: Integrates evolutionary information derived from sequence profiles.
• Position-specific scoring matrices (PSSMs): Uses PSSMs to represent evolutionary conservation of residues.
• Mono-gram and bi-gram profiles: Extracts mono-gram and bi-gram features from PSSMs for model input.
• 1D-convolutional neural network: Employs a 1D-CNN to learn spatial hierarchies in sequence-derived features.
• Performance improvement: Demonstrated an enhancement in prediction accuracies by more than 4% on independent test datasets.

Scientific Applications:

• Protein function classification: Identification and classification of DNA-binding proteins as SSBs or DSBs.
• Functional inference in DNA metabolism: Support studies of proteins involved in DNA replication, packing, and repair by providing binding-type annotations.

Methodology:

Evolutionary information is extracted via position-specific scoring matrices (PSSMs) to compute mono-gram and bi-gram profiles, which are used as input to a 1D-convolutional neural network classifier.