Visualizing the gene network obtained by WGCNA

Question

[seta]

Dear all,

I'm learning WGCNA via following its tutorial and example (Network analysis of liver expression data in female mice). However, I came across some questions, happy to know your comments.

Regarding visualizing the gene network, the tutorial says:

# Calculate topological overlap anew: this could be done more efficiently by saving the TOM
# calculated during module detection, but let us do it again here.
dissTOM = 1-TOMsimilarityFromExpr(datExpr, power = 6);
# Transform dissTOM with a power to make moderately strong connections more visible in the heatmap
plotTOM = dissTOM^7;
# Set diagonal to NA for a nicer plot
diag(plotTOM) = NA;
# Call the plot function
sizeGrWindow(9,9)
TOMplot(plotTOM, geneTree, moduleColors, main = "Network heatmap plot, all genes")

Here, my question is while the power was determined 6 (power=6) in the previous line, why the power of 7 was used in dissTOM^7?

Also, my obtained heatmap had a red background. I solved this issue using this post. ,

however, as you may see in the above image, the plot colors are not identical to the corresponding plot in the tutorial. I would like to know if it is the matter, I mean does it impact the results?

Thanks

[andres.firrincieli]

Hi seta

Here, my question is while the power was determined 6 (power=6) in the previous line, why the power of 7 was used in dissTOM^7?

I do not think there is any specific reason, 6 or 7 should not make too much difference. It could be just an error in the code. In this case, the only purpose of raising the dissTOM to a power of 7 (or 6) is to enhance the contrast between gene pairs with low dissimilarity (red) from those with high dissimilarity (Light orange)

Answer 1

[LauferVA]

The answer is that while the first toplogical overlap matrix (TOM) assigns an exponent in order to create a soft threshold (Langfelder 2008), the second exponent is an arbitrary selection designed to accentuate the strongest networks in the data at the expense of the other relationships.

While the soft thresholding procedure itself is more art than science, there is some justification provided for it. See, for example, this post: how is the topological overlap (TOM) algorithm calculated?. By contrast, the second transform is used purely for visualization. The reason why it aids with visualization is that, as you raise numbers between 0 and 1 to a high exponent, everything will rapidly approach 0 unless the starting connection is very strong.

Let me know if you need further clarification.

[seta]

Thank you very much for the comments. Just one question related to this part; the eigengene dendrogram show groups of correlated modules. Could you please let me know how I can have the correlation coefficient of different modules?

[andres.firrincieli]

here is the tutorial

[seta]

Thanks. I missed it.