Plant protein interaction network constructing method based on deep learning

Abstract

The invention relates to a plan protein interaction network constructing method based on deep learning. The method comprises the following steps of 1), acquiring 11 characteristic data of a protein interaction pair; 2), performing screening for obtaining a training set and a testing set; 3), constructing a deep learning classification model; 4), performing batch optimization on the parameter of the deep learning classification model, and obtaining a classification model of an optimal optimization parameter combination; 5), performing interaction relation predicting on all possible two-to-two interaction proteins of the whole genome according the classification model of the optimal optimization parameter combination; and 6), predicting the protein interaction network according to an interaction relation predicting result. Compared with the prior art, the method has advantages of high predicting accuracy, high modeling efficiency, etc.

Description

technical field [0001] The invention relates to deep learning technology in the field of biotechnology, in particular to a method for building a plant protein interaction network based on deep learning. Background technique [0002] Protein interaction is essential in the biological process of cells, and most genes perform their biological functions by interacting with other proteins at the protein level. The advent of the post-genome era has provided abundant data information for predicting protein interactions on a genome-wide scale, and with the development of high-throughput experimental technology and bioinformatics, the research progress of complex biological networks has been greatly improved. [0003] Building a classification model requires the use of statistics, machine learning, and other methods to extract valuable information from a large amount of data. This process includes data preprocessing, classification, and anomaly detection. With the explosive growth o...

AI Technical Summary

Problems solved by technology

However, most studies on the construction of plant-protein interaction networks use traditional machine learning methods such as decision trees, naive Bayesian, support vector machines, and random forests for modeling, and few studies use deep learning methods to build protein interaction classification models. very few, greatly limiting the possibility of improving forecast accuracy

[0004] In addition, the deep learning model has a large number of parameters and requires a lot of training data, resulting in a complex model, a large amount of calculation, and easy to cause training overfitting

Images