Build an artificial intelligence framework to find the most effective lysozyme
Author：Eve Cole
Update Time：2024-11-22 19:30:01
The reporter learned on September 2 that the team of Lin Zhanglin, a professor at the School of Biomedicine at Guangdong University of Technology, and Yang Xiaofeng, an associate professor at the School of Biology at South China University of Technology, successfully built a new artificial intelligence framework - DeepMineLys, and used this framework to discover the By far the most potent lysozyme in the human microbiome. The research results were recently published in Cell Reports. "This study uses artificial intelligence to identify and mine lysozyme with the potential to treat drug-resistant bacteria from human microbial metagenomic data. This is an important breakthrough in the use of artificial intelligence in biological research." First author of the paper, South China Fu Yiran, a 2018 doctoral student from the School of Biological Sciences and Engineering of the Polytechnic University, said. Lin Zhanglin believes that the success of DeepMineLys is due to the construction of a comprehensive training data set covering a wide range of lysozymes, the integration of advanced algorithms and coding technologies such as TAPE, and the adoption of several key factors such as a three-layer convolutional neural network and a dual-track architecture. Greatly improved prediction performance. The research team randomly selected 16 lysozymes from the top 100 candidate lysozymes for experimental verification, 11 of which were confirmed to be active. The activity of the strongest lysozyme was even 6.2 times higher than that of traditional lysozyme, making it the most active lysozyme in the human microbiome so far. The most potent lysozyme found in . "DeepMineLys can not only mine lysozyme, but also has the potential to mine proteins. It can be seen that the construction of this framework provides a powerful tool for future biological research." Lin Zhanglin said.
