I'm exploring WGCNA for bulkRNA sequencing analysis with human subjects. I have healthy, myocarditis, and heart failure patients (which can be further broken into ischemic or nonischemic)

I have been spending quite a bit of time tinkering with the parameters in blockwisemodules() to identify modules/MEs, and I'm only able to assess the quality of this step with a rudimentary intuition.

For example, if I use

net = blockwiseModules(datExpr, power = 10,maxBlockSize = 20000,TOMType = "unsigned", minModuleSize = 20,reassignThreshold = 50, mergeCutHeight = 0.35,numericLabels = TRUE, pamStage = TRUE, pamRespectsDendro = TRUE)

I get what is so far the "best" module detection - many modules (although small), and about 60-70% of genes seem to fall into a module. Image below.

I find myself at a bit of a loss as to how to judge the quality of this step. The module detection looks extremely messy compared with the tutorial data, which i wasn't surprised at really, but can anyone provide guidance here?

I can provide other images of what I get from this step that looks like much poorer detection of modules if it would help.

Last related question, I have 19 healthy, and about 41 patient samples, I wasn't sure if the best approach to module detection is to put all samples together, then look at what modules are associated with disease states which i could convert into numeric for that step, or would it be better as i have started - to detect modules in healthy, or patient groups in independent datasets?

enter image description here



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.