Project name: Decrypting Antimicrobial Proteins from the Gut Virome Using Deep Learning

Description: Despite the promise of phage therapy, targeting gut bacteria, especially those linked to chronic diseases, remains challenging. Phage lysins, with their antimicrobial properties and ease of synthesis, offer a promising alternative, yet current lysin databases focus on clinical pathogens. We introduce VirHost Hunter, a deep learning framework integrating features from tail and lysin proteins. Trained on diverse datasets and validated with experimental evidence, excels in predicting phage hosts with high precision and sensitivity across various taxonomic levels. To broaden VirHost Hunter's application, we calibrated VirHost Hunter for chronic disease-associated gut bacteria. We established the Gut Phage Lysin Database (GPLD), which encompasses 117,698 lysins precisely targeting 29 disease-related gut bacterial families, enhancing lysin engineering and therapeutic use. As a proof of concept, we identified and synthesized a lysin from our database, confirming its efficacy and specificity against an obesity-associated bacterium. VirHost Hunter advances our understanding of phage-host interactions and precision therapies for microbiome-related disorders.

Data type: Protein sequence

Sample scope: Multispecies

Relevance: Other

Release date: 2026-02-24

Last updated: 2026-02-24

Data size: 60MB