What is `soft thresholding` with WGCNA?

Question

[O.rka]

I have been using the WGCNA package (https://labs.genetics.ucla.edu/horvath/CoexpressionNetwork/Rpackages/WGCNA/faq.html) for correlation networks.
http://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-9-559

I understand what a hard threshold is: absolute value of correlation matrix, choose a cutoff (e.g. 0.85), anything above is considered connected in the network. But then there is soft thresholding which is when you exponentiate the correlation matrix and that accentuates larger connections.

How do you then decide which ones are connected or not? Do you do a hard threshold after the soft thresholding?

Answer 1

[sina.nassiri]

In a weighted network (as is the case in WGCNA) all genes (nodes) are connected to one another. The idea behind soft thresholding (raising similarity measures to a power) is to emphasize more on stronger associations (larger correlation coefficients). The intuitive logic behind soft thresholding is that in reality the interactions are never black and white (1s and 0s). Therefore, in a weighted network you keep all pairwise interactions and take them into account in your analyses. This will better capture the continuous nature of pairwise interactions in the system. That being said, even when dealing with weighted networks you may end up applying some sort of arbitrary hard threshold on the edge weights for plotting purposes. This is simply because keeping all interactions makes your network way too messy to visualize.