PTBGRP

PTBGRP predicts bacteriophage-bacteria interactions to identify potential phage candidates for treating bacterial infections.

Key Features:

• Microbial Heterogeneous Interaction Network (MHIN): Integrates phage-bacteria interactions and six bacteria-bacteria interaction networks to represent multi-relational microbial relationships.
• Representation Learning: Extracts high-level and low-level features from the MHIN to encode node and network patterns for prediction.
• Deep Neural Network Classifier: Applies a deep neural network to classify phage-bacteria interaction pairs based on learned representations.

Scientific Applications:

• Benchmark evaluation on ESKAPE and PBI datasets: Demonstrates improved predictive performance on datasets involving ESKAPE pathogens and PBI.
• Case studies on specific pathogens: Provides predictions for clinically relevant bacteria such as Klebsiella pneumoniae and Staphylococcus aureus.
• Phage therapy research: Supports selection of candidate bacteriophages for development of targeted treatments against resistant bacteria.

Methodology:

Constructs a Microbial Heterogeneous Interaction Network (MHIN) integrating phage-bacteria and six bacteria-bacteria interaction networks, applies representation learning to extract high- and low-level features from the MHIN, and classifies interactions using a deep neural network.