Feature gene selection method based on deep learning and evolutionary computation

Abstract

A feature gene search method based on deep learning and evolutionary computation includes the following steps: 1) calculating a differential expression level index and establishing a primary gene poolaccording to a Pareto principle; 2) according to the expression level of each gene, calculating a density matrix and a distance matrix after the mapping of the genes, drawing a decision map, using multiple linear regression analysis to fit a binary plane, and automatically determining a clustering center; 3) constructing a deep gene expression prediction network to calculate gene-genetic susceptibility information (GGSI) of a primary gene pool; 4) removing redundant genes based on GGSI values and creating a secondary gene pool; 5) performing binary coding on a cuckoo search algorithm based onthe GGSI values, selecting the most compact gene sets, and establishing a tertiary gene pool. According to the method, a feature gene selection framework is provided based on a hierarchical structure, which can extract key genes better and use multiple linear regression analysis combined with a deep learning algorithm and an optimization algorithm to select the most compact feature gene sets.

Description

technical field [0001] The invention belongs to the field of biological information, and in particular relates to a method for selecting characteristic genes. Background technique [0002] With the advancement of gene sequencing technology, high-throughput sequencing technology provides a large amount of gene expression data. Gene expression profiling is widely used as a tool for capturing cellular expression patterns following disease occurrence, genetic perturbation, and drug treatment. Facing a large amount of high-dimensional gene expression data, how to mine useful information has become the research focus in the field of bioinformatics. [0003] Due to the high-dimensional nature of gene expression data, mining characteristic genes with key information is of great significance for subsequent gene data processing and sample phenotype analysis. At present, some researchers combine binary particle swarm optimization algorithm (BPSO) and filtering method to find the best...

AI Technical Summary

Problems solved by technology

However, these hybrid methods lack interpretability while simplifying the gene set and achieving high processing efficiency.

Images