I have 2 coexpression networks: (1) control phenotype; and (2) treatment phenotype. They are using the same subset of genes so they are directly comparable (n=1700). I used WGCNA to create 2 separate coexpression networks using the same soft power threshold (beta=10) for both. This question is more conceptual than actual commands or tools to use but I am looking for methods that can allow me to compare the differences in network topology between the 2 phenotypes. My first idea was to calculate the differential coexpression network as suggested in this paper https://www.nature.com/articles/srep13295 but it's a little weird to work with the resulting matrix. The only other approach I could think of was to have a pairwise analysis of modules between each network to get an overlap measure.
Does anyone know of any methods, tools, or approaches that work well for datasets like this?
My main research question is to identify the biggest changes in network topology between the 2 phenotypes.
It's a dataset with 49 samples of patients without a disease and 34 with a disease. 28 of the patients have a twin in the dataset (n=28*2=56) and 27 do not have a twin account for in the dataset for a total of 56 + 27 = 83 samples/patients. The gene subset I'm investigating is 1700 so each network consist of a 1700x1700 symmetrical similarity matrix. There is a separate network for the healthy state and a separate one for the diseased state.