What data should I use to generate a gene coexpression network?

Question

[tahermun]

I have a microarray dataset that contains expression data of 30 samples of individuals with a certain disease and 30 samples of healthy individuals. After restricting this data to the most significantly differentially expressed genes, I want to use the WCGNA R package to perform a gene coexpression network analysis. Ideally, I want to look at the coexpression network of the disease data and the coexpression network of the healthy data separately, so I was thinking about separating the data and running WGCNA separately on both types of samples. Would this give me bad results? Do I need multiple types of samples to construct a correlation network?

Essentially, my question is: do I need different classes of samples to perform one coexpression network analysis, or could I construct a network for each class separately?

Answer 1

[Deepak Tanwar]

I was thinking about separating the data and running WGCNA separately on both types of samples. Would this give me bad results?

I don't think so.

question is: do I need different classes of samples to perform one coexpression network analysis, or could I construct a network for each class separately?

In principle, you should construct the coexpression networks separately. Then only you would be able to compare the gene networks.