Is using WGCNA package for analyzing 12 samples in 3 studies possible?

Question

[DrEruhami]

Hello there people I'm having trouble with using the WGCNA package. Does anyone know the least amount of samples for running this package? Mine are 12, would it be enough? This is the code in case of being useful

>install.packages("BiocManager")
>BiocManager::install("WGCNA")

>install.packages(c("matrixStats", "Hmisc", "splines", "foreach", "doParallel", "fastcluster", "dynamicTreeCut", "survival"))
>source("http://bioconductor.org/biocLite.R")
>biocLite(c("GO.db", "preprocessCore", "impute"))

[ATpoint]

The developers (in the FAQ section of WGCNA) clearly state that they at least recommend 15 samples, better 20 or more so no, 12 would not be ok, given that you put trust in the developers expertise.

[Kevin Blighe]

12 is okay

[ATpoint]

Users are free to do as they like, this is what the developers recommend though:

We do not recommend attempting WGCNA on a data set consisting of fewer than 15 samples. In a typical high-throughput setting, correlations on fewer than 15 samples will simply be too noisy for the network to be biologically meaningful. If at all possible, one should have at least 20 samples; as with any analysis methods, more samples usually lead to more robust and refined results.

https://horvath.genetics.ucla.edu/html/CoexpressionNetwork/Rpackages/WGCNA/faq.html

[DrEruhami]

Thanks, guys your answers are truly helpful, I will consider them. Thank you Kevin

[Kevin Blighe]

You are very welcome

Answer 1

[Kevin Blighe]

12 samples is okay. However, you imply that these 12 samples are split across 3 studies, in which case you should anticipate batch and other study-design differences. Please check.

By the way, please do not install packages via source() and bioclite() - this method is deprecated. Please instead use BiocManager::install() or just install.packages().

Kevin