Applications

• Targeted phage therapies for antibiotic-resistant infections
• Personalized medicine through rapid phage-host prediction
• Designed phage combinations for broad-spectrum and biofilm-related infections
• Biocontrol and environmental cleanup using engineered phages

Advantages/Benefits

• Predicts phage receptor  directly from genome data
• Enables high-throughput screening with AUROC > 0.9 for most targets
• Streamlines the phage-host matching process and reduces resource use
• Scales to thousands of phage genomes across diverse species

Background

Phages are powerful tools for treating antibiotic-resistant infections and engineering microbial communities, but identifying which phages infect which bacteria remains a major challenge. Traditional approaches rely on sequence similarity and time-consuming lab work, making them difficult and less reliable for broad application. This technology addresses that challenge by using machine learning trained on genome-wide genetic screens to predict phage-host adsorption factors directly from phage genome sequences, enabling rapid, high-throughput identification of phage targets.

Technology Overview

Berkeley Lab researchers have developed a machine-learning platform that predicts which receptors bacteriophages use to infect their bacterial hosts. The model is trained on genome-wide genetic screens (RB-TnSeq, Dub-seq, CRISPRi) and phage genome sequences. It uses protein k-mer signatures and a step-by-step filtering process to identify the most important features linked to phage receptor-binding proteins and their target receptors. The system was tested on 180 phages and two E. coli host strains, achieving high accuracy, with AUROC values above 0.9 for most tested receptors. Unlike traditional methods that rely on genetic similarity, this approach identifies functional binding sites across thousands of phage genomes in just minutes. This technology enables faster, large-scale phage therapy design, receptor discovery, environmental applications, and synthetic biology research.

Development Stage

Proof of Concept 

For More Information:

N/A

Principal Investigator

• Lucas Moriniere
• Avery Noonan 
• Adam P Arkin
• Vivek K Mutalik 

Status

Patent pending 

Opportunities

Available for licensing or collaborative research