I am working on a project trying to extract features from RNAseq data from monkeys that were challenged with Ebola Virus, and to build a classifier that could predict the disease stages of Ebola infection. During the process of literature research, I came across your the Bioconductor package Pigengene, which almost fits my needs perfectly.
However, as I was trying to use the main function, one.step.pigengene(), directly, there showed an error message "power is NA!". After I studied the source code, I found the issues lied in the calculate.beta() function, were a soft-threshold power should be picked to fit a scale free network. I tried to run each step from the WGCNA package sequentially (following your package), while feeding in a random power term, and everything worked fine, even though the results were not satisfying (obviously the net work was not optimized). I started to suspect this might be caused by the data I tried to analyzed (the number of genes included vs. the number of data point).
For information, I have 15 monkeys, with 4 time points each (so total of around 60 datapoints from RNAseq), and I fed around 4000 genes that were significantly differentially expressed into the package.
Any suggestions or help would be really appreciate!!!
Thank you so much!!!