Is it proper to compare two WGCNA matrixes if they were raised to different power (soft threshold)?

Question

[cwwong13]

This is a theoretical question. I am trying to do a comparative analysis (meta-analysis, preservation analysis, differential analysis, etc.) on the mRNA expression network constructed from WGCNA. I acknowledge that we were advised to choose the least power during the "pick soft threshold" step. However, essentially, the networks I am trying to compare were using different power.

My question is: What would be an appropriate procedure to compare multiple networks?

Does it matter if the networks are transformed by different power during the soft thresholding step? Similarly, does it matter if the networks/ modules were defined with different "deep" during the cut tree step?

I really appreciate your help.

Answer 1

[andres.firrincieli]

In WGCNA the module preservation analysis does not require two networks but one network (reference network) and a test dataset (normalized expression data). The analysis will tell you if the modules in the reference network are reproducible in the test dataset.

This approach is explained here: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1001057