WGCNA - higher number of modules

Question

[marina.wakid]

Hello all,

I'm currently performing WGCNA with the following parameters:

network <- blockwiseModules(counts.wgcna, power= 14, corType="pearson", maxBlockSize = 30000, networkType="signed", minModuleSize = 50, nThreads=0, TOMType = "signed", TOMDenom = "min", deepSplit=2, verbose=5, mergeCutHeight=0.15, reassignThreshold = 1e-6, detectCutHeight = 0.995, numericLabels=TRUE, saveTOMs=FALSE, pamStage=TRUE, pamRespectsDendro=FALSE, randomSeed = 54321)

I believe I have performed every step correctly, however I'm getting 201 modules as an output. I know that this is particularly high and that I could change the deepSplit and minModuleSize, however I'm concerned that the output might become more generic if I do so. How many modules is deemed sensible? i.e. that implies a good compromise between resolution and network cutting/module detection?

Thanks!

[allingt]

Have you filtered low expression genes? How did you normalise the data?