In FAQ's of WGCNA package it's mentioned "We do not recommend attempting WGCNA on a data set consisting of fewer than 15 samples". What could be an alternate approach if one have let's say 6 samples? Can anybody explain if for example I want to analyze this gene expression data then from where will I start?
You can make an easy co-expression network by simply constructing the symmetric correlation matrix of all genes (adjaceny matrix as its called in WGCNA) then filtering that matrix by a hard threshold such as r = 0.70.
Although, based on your newest comment i am not sure why differential expression analysis is not sufficient. Couldn't they be ranked on highest absolute fold change or lowest FDR adjusted p-value?