CoMPHI: A Composite Machine Learning Approach to Predict Hosts of Bacteriophages

Overview

Phage therapy has reemerged as a compelling alternative to antibiotics in treating bacterial infections, especially antibiotic-resistant superbugs. The challenge in the broader application of phage therapy in modern medicine is identifying host targets for the vast array of uncharacterized phages. To solve this issue, this paper introduces an innovative composite machine-learning model designed for predicting phage-host interactions by combining the accuracy of alignment-based methods with the efficiency and flexibility of machine-learning techniques. This model is called CoMPHI (Composite Model to predict Phage Host Interaction). The model initially generates multiple feature encodings from nucleotide and protein sequences of phages and hosts to enhance prediction accuracies. It is further enriched by incorporating alignment scores between phage-phage, phage-host, and host-host, creating a composite model. During the evaluation using 5-fold validation, the composite model exhibited an AUC of 94% and an accuracy of 92.3% at the species level, increasing to 96.7% and 95.1% at the phylum level, respectively. A comparative analysis revealed a 6-8% increase in model performance due to the inclusion of alignment scores. Additionally, an ablation study highlighted that including nucleotide and protein sequences from both phages and hosts increased the model’s performance. Another ablation study provided evidence that alignment scores from phage-host and host-host, combined with phage-phage scores, equally enhanced the model's performance. The applications of the model are varied, ranging from healthcare to agriculture to the environment. In conclusion, this paper presents a robust and comprehensive composite model advancing the widespread use of phage therapy in modern medicine.

Competition history

• AJAS 2025 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• AJAS None projects
• Every AJAS 2025 project