As harmful bacteria become more resistant to antibiotics, phage therapy is expected to be the ultimate weapon against bacterial resistance.
However, in the actual application process, phage therapy is difficult to be widely promoted around the world, and the main reason is the specificity of phages. Most phages have very specific interactions with their bacterial hosts, which makes it difficult to find one or a few phages that match a particular pathogen and does not scale well to screen hundreds or even thousands of phages.
So the experts asked the question: Can we develop a computational tool to screen phages in computers in a practically relevant way? The answer is: AI technology!
Application of Artificial Intelligence in Bacteriophage Research
In recent years, artificial intelligence (AI) technology has been widely used in the biomedical field, and important progress has been made in phage and host genotyping.
01AI-assisted phage screening and identification
By analyzing large amounts of phage and host genome data, AI models can quickly and accurately predict and identify potential phage-host pairing relationships. The deep learning-based model can extract key features from the phage genome sequence and match them with the genetic markers of the host bacteria to achieve automated, high-throughput phage host recognition.
02 The Optimization Role of Artificial Intelligence in the Application of Bacteriophage Therapy
By integrating multi-source information such as phage genome, host mechanism, and clinical data, AI models can predict the ability of specific phages to infect target bacteria, as well as their kinetics in vivo, providing important support for the design of phage treatment regimens.
An article published in Nature communications on May 22 this year proposes a machine learning system called PhageHostLearn that can predict strain-level interactions between Klebsiella phage-bacteria receptor-binding proteins and bacterial receptors, providing a framework for the development and evaluation of phage host prediction methods.
03Discover novel lyases using artificial intelligence
Recently, the team of Professor Li Jinquan of Huazhong Agricultural University published a research paper in the international journal Advanced Science. The research released "DeepLysin", a one-stop lyase discovery and activity evaluation software, which can use artificial intelligence to identify the nuances of high-dimensional signatures of antimicrobial candidate proteins, enabling efficient screening of lyase candidates from "dark matter" such as phages or metagenomes. This lyase is a class of protein-based antimicrobial substances, which has the advantages of efficient and rapid bactericidal effect and is not easy to induce bacterial resistance.
（DeepLysin工作流程示意图）
In short, in the study of bacteriophages, AI technology is the outlet, which can greatly help experts and scholars solve some problems that were difficult to solve before.