selection soft Power in WGCNA

Question

[rahmati.razieh83]

Hi everyone I'm running WGCNA pipeline for 18 samples with about 60000 genes. I do not know which soft power I should consider regarding the plots. As scale independence plot shows two section I do not know whether 4 or 19 should be considered as soft power. Please guide me.

Answer 1

[andres.firrincieli]

I think you are working with transcripts and not genes; 60000 seems a little to much. By the way, did you check the data for the presence of outlier samples? Did you filter out low count transcripts?

[rahmati.razieh83]

Dear Andres Thanks for your answer I am new in WGCNA, so I do not how to perform WGCNA on my data. But as you answer to my question, I reduced the number of contigs I got from RNA Seq. I had a transcriptome data with 303000 contigs and after removing the redundant transcripts by Cd-Hit_EST and filtering out the low count transcripts, the number of contigs reduced to 60000 contigs. I selected the power of 19 and run the WGCNA on a system with 2 Tb RAM. After one week of running, WGCNA was crashed and I got the bellow error. Please guide me about this error and let me know how can I run WGCNA with this too much data

TOMplot(dissTOM , geneTree,dynamicColors, terrainColors=TRUE)

* caught segfault * address 0x7f0f6dae726c, cause 'memory not mapped'

Traceback: 1: image.default(x = 1:nc, y = 1:nr, x, xlim = 0.5 + c(0, nc), ylim = 0.5 + c(0, nr), axes = FALSE, xlab = "", ylab = "", ...) 2: image(x = 1:nc, y = 1:nr, x, xlim = 0.5 + c(0, nc), ylim = 0.5 + c(0, nr), axes = FALSE, xlab = "", ylab = "", ...) 3: .heatmap(as.matrix(dissim), Rowv = dendro, Colv = dendro, scale = "none", revC = TRUE, ColSideColors = as.character(labeltree), RowSideColors = as.character$ 4: TOMplot(dissTOM, geneTree, dynamicColors, terrainColors = TRUE) An irrecoverable exception occurred. R is aborting now ...

[andres.firrincieli]

I never encountered that error. If you google 'caught segfault memory not mapped' you will see that the error is caused by different issues. Whetere or not is worth trying to fix the error and wait days to see if it is fixed, is up to you. Personally, I would not waste time on creating a plot of the Topological Overlap Matrix (TOM) since it is not that informative.

You have everything you need, i.e., adjacency matrix, TOM and modules (dynamicColors) to perform a very basic analysis with WGCNA: calculate the Module Eigengenes (moduleEigengenes()) to find biologically meaningful modules, calculate the intramodular connectivity to find the hub genes, use the TOM to viasualize the modules in Cytoscape.

I'm sorry I couldn't help you with the error.

Good luck!

Answer 2

[tujuchuanli]

Thank you for your wonderful discussion!