iProm-phage

iProm-phage predicts and classifies promoters in bacteriophage genomes to support accurate annotation and analysis of phage regulatory regions.

Key Features:

• Two-Layer Model Architecture: Employs a two-layer model where the first layer discriminates promoter versus non-promoter sequences and the second layer classifies identified promoters as phage-specific or host-derived.
• Enhanced Negative Dataset Creation: Constructs a challenging negative dataset by using promoter sequences as negatives rather than non-coding regions to improve discrimination between true and false positives.
• Feature Encoding and Machine Learning: Evaluates 10 feature encoding methods, selects one-hot encoding, and applies a 1-D convolutional neural network (CNN) for sequence pattern recognition.
• Performance and Validation: Assesses predictive performance using rigorous 5-fold cross-validation, reporting high predictive accuracy.

Scientific Applications:

• Phage Genome Annotation: Identifies and classifies promoters within phage genomes to facilitate detailed annotation and study of phage regulatory elements.
• Antibacterial Research: Provides insights into phage promoter regions relevant to research on bacteriophages as antibacterial agents and regulation of phage gene expression.

Methodology:

Implements a two-layer classification framework using one-hot encoded sequences, evaluates 10 encoding methods, trains a 1-D CNN with a negative dataset composed of promoter sequences, and validates performance with 5-fold cross-validation.