Width learning zymoprotein detection method and system based on global sampling subgraph

Abstract

A width learning zymoprotein detection method based on a global sampling subgraph comprises the following steps: S0, structure conversion: converting a protein molecular structure into a graph structure; S1, global sampling: carrying out N times of global sampling on an original image according to connecting edges to obtain N sub-networks; S2, sub-graph mapping: performing first-order and second-order mapping on the N sub-networks according to an SGN rule to obtain 2N mapping networks; S3, feature extraction and feature fusion: conducting feature extraction on the original network and the 2N mapping networks based on Graph2vec to obtain K-dimensional network representation vectors of 2N+1 networks, and obtaining (2N+1)*K-dimensional feature vectors through transverse splicing of the representation vectors to serve as final representation of the original network; and S4, training of a width network classifier: training the width network through the final representation of the original network and the supervised network label, and finally obtaining the detection precision of the zymoprotein through ten-fold cross validation. According to the invention, efficient and accurate zymoprotein detection is realized.

Description

technical field [0001] The present invention relates to network science, data mining and enzyme and protein detection technology, in particular to a method and system for detecting enzyme and protein of breadth based on global sampling subgraph. Background technique [0002] In recent years, graph data has received more and more attention. In real life, social network, biological protein network, and literature citation network can all be characterized by graphs. Enzyme proteins exist in various forms in nature, and how to identify whether a protein is enzymatic has broad prospects in the field of biocatalysis. The problem of graph classification is a common task in graph data mining. For example, it is widely used in protein toxicity inference and chemical molecular property prediction. Therefore, it becomes very useful to combine graph classification and enzyme detection. significance. [0003] A subgraph is a basic component in the network, which can be used to describ...

AI Technical Summary

Problems solved by technology

This method uses a random walk strategy to obtain a local network structure, but the lack of global intrinsic information leads to a lack of classification accuracy, and the use of an extreme random tree as a classifier still needs to be enhanced in terms of classification training speed

Images