Design for gene network analysis with WGCNA

Question

[yh362]

I have lots of RNA-seq data from various tissues and conditions, etc of a polyploid plant, and wanna take a look at how the two subgenomes behave in terms of co-expression network. Can I pool all these RNA-seq data together? Or is it more reasonable to split them by tissues and conditions and construct co-expression network separately?

Answer 1

[andres.firrincieli]

How many tissues and conditions do you have?

Pooling the data or split them by tissues and conditions is up to you. Pooling the data all together should give you an idea of how genes are co-expressed across tissues and conditions; by looking at the modules eigengenes you can find e.g., tissue-specific modules, condition-specific modules, tissues x conditions-specific modules

For a differential network analysis you have to split the data by tissues or conditions depending on what you are looking for. For example, if you are interested in how the module structure and connectivity change across conditions then you have to build a co-expression netowork for each condition and compare the networks against each other eg. NetworkCond1 vs NetworkCond2; NetworkCond2 vs NetworkCond1 NetworkCond1 vs NetworkControl; NetworkCond2 vs NetworkControl etc... in this case you need at least 15 samples for each condition.