I am a beginner in bioinformatics, currently learning about Weighted Gene Co-expression Network Analysis (WGCNA). I understand that WGCNA is a powerful method for constructing gene networks, detecting gene modules, and identifying hub genes within these modules. However, I am confused about one specific aspect of WGCNA: the use of weighted correlation for network construction instead of directly using Pearson correlation. From my understanding, Pearson correlation measures the linear relationship between two variables, and it is commonly used in many statistical analyses, including gene expression studies. Given its simplicity and direct interpretation, I am curious why WGCNA prefers a weighted approach. Could anyone explain the advantages of using weighted correlation in WGCNA over the direct application of Pearson correlation? Thank you in advance.
The WGCNA implementation by Langfelder and Horvath provides various correlation methods to create the initial correlation matrix. I don't think any of them are weighted. The W in WGCNA means that the networks (adjacency matrix) are weighted, not the correlations. Check out their article https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-9-559 and the WGCNA R package manual.