A prediction method of lactic acid bacteria antimicrobial peptides based on graph neural network

Abstract

The invention discloses a lactic acid bacteria antibacterial peptide prediction method based on a graph neural network. A positive sample is established by searching for known lactic acid bacteria antibacterial peptides, and a negative sample is established by collecting sequences with a length of 5‑255 from a protein database. Redundant sequences are removed and similar ; According to the feature extraction of positive and negative samples, the feature vector and the initial input image are obtained, and the graph neural network model is established on this basis; through the training, evaluation and cycle optimization of the graph neural network model, the optimal number of layers of the graph neural network, the maximum parameters such as the optimal number of training rounds and learning rate; finally, the data of strains suspected to have antibacterial activity were predicted based on the graph neural network model. The present invention adopts the above-mentioned lactic acid bacteria antibacterial peptide prediction method, replaces laboratory wet experiment screening with computer model prediction, shortens the judgment time of lactic acid bacteria antibacterial peptide protein sequences, realizes accurate and efficient batch identification, and provides an effective alternative for the screening of lactic acid bacteria strains with antibacterial properties method.

Description

technical field [0001] The present invention relates to the identification field of biological antibacterial peptides, in particular to a method for predicting lactic acid bacteria antibacterial peptides based on graph neural network. Background technique [0002] In the existing identification technologies of bio-antimicrobial peptides, the following two technologies are mainly used: [0003] 1. The bacteriostatic experiment using the agar hole diffusion method is time-consuming and cannot be identified by high-throughput; 2. The identification is carried out by using machine learning technology or long-term memory and convolutional neural network technology in deep learning, although it can process multiple The amino acid sequence can only capture the local semantic information of the antimicrobial peptide sequence, and it is not easy to grasp the characteristic information of the antimicrobial peptide from the perspective of the overall structure, so the recognition accur...

AI Technical Summary

Problems solved by technology

[0003] 1. The antibacterial experiment using the agar hole diffusion method takes a long time and cannot be identified at high throughput; 2. The identification is performed using machine learning technology or long-term short-term memory and convolutional neural network technology in deep learning, although multiple Amino acid sequence, but it can only capture the local semantic information of the antimicrobial peptide sequence, and it is not easy to grasp the characteristic information of the antimicrobial peptide from the perspective of the overall structure, so the recognition accuracy and other indicators need to be improved

Images