Subgroup specific co-expression network identification method

Abstract

The invention discloses a subgroup specific co-expression network identification method. The invention provides a method for identifying a specific co-expression network of a subgroup of a sample to be detected. The method comprises the following steps of: 1) carrying out homogenization treatment on an original UMI number value of each gene of each sample in the subgroup of the sample to be detected; 2) filtering an original UMI homogenization result; 3) calculating the weighted median correlation bicor of every two filtered genes; 4) filtering the weighted median correlation result to obtaina strong co-expression gene relationship pair; 5) subtracting the shared strong co-expression gene relationship pair appearing in all cells in the subgroup from the filtered strong co-expression generelationship pair of the subgroup of the cells to be detected to obtain a specific co-expressed gene pair of the subgroup of the cells; and constructing a specific gene co-expression network of the subgroup of the sample to be detected by using the specific co-expression gene pair of the subgroup of the cells. The method for the subgroups is more targeted and more sensitive.

Description

technical field [0001] The invention belongs to the field of biotechnology, and in particular relates to a subgroup-specific co-expression network identification method. Background technique [0002] With the development of sequencing technology and the decline of sequencing costs, the sample size for gene expression research is increasing, especially the introduction of 10x single-cell RNA-Seq technology, which makes the number of samples for gene expression research jump to a thousand. For expression The study of differential changes has changed from simple sample-to-sample and group-to-group comparisons to clustering of subgroups of samples based on gene expression changes, and then looking for specifically expressed genes based on the results of the subgroups , or differentially expressed genes, to explain the functions performed by different subgroups of cells, or the reasons why individuals of different subgroups have differences. [0003] The current method used to f...

AI Technical Summary

Problems solved by technology

[0005] 1. It is impossible to guarantee that all cells / samples are in the same state at the same time in one sampling, and the gene expression level is not necessarily consistent. Moreover, due to the low initial amount of single-cell data, there will often be some loss of expressed genes. It is simple, crude and inaccurate to take the average of gene expression in subgroups for analysis;

[0006] 2. By selecting genes with significant differences, only target genes with particularly obvious differences can be found, but regulatory genes with less drastic changes cannot be identified, which has limited effect on understanding the function of subgroups or explaining the reasons for phenotypic differences

[0007] 3. The regulation of gene expression in organisms is real-time and dynamic, and the realization of most functions is not achieved by drastic changes in one or several independent genes, but is regulated by a group of genes synergistically. The degree of change of the gene alone may not be significant, and it cannot pass the screening of the multiple of difference and p-value

[0009] 1. This method is only effective when the number of samples is small (a dozen to one hundred). When the number of samples increases, the consumption of memory and time increases sharply; in addition, the number of samples (cells) of 10x single-cell data reaches thousands level, sometimes even exceeding the number of expressed genes, at this time, the WGCNA algorithm is completely inapplicable

Images