Soft Thresold - WGCNA

Question

[Silvia Gonzalez]

Hi, I have done a WGCNA with my data. Now I want to know some technical concepts that I put in my WGCNA code.

First of all, I have seen that in general people put their powers of SoftThresold between these values:

powers = c(c(1:10), seq(from = 12, to=20, by=2))

Why at first we put 1:10 in a scale of one by one and then when we arrive at 12 we start to put the range two by two?

Second, what it means the “verbose”? How can I know which verbose it’s the better for my data?

sft = pickSoftThreshold(DatExprs, powerVector = powers, verbose = 5)

Third question, when we are doing the Scale Independence plot we put the following code:

plot(sft$fitIndices[,1], -sign(sft$fitIndices[,3])*sft$fitIndices[,2]………….

In the function fitIndices we put [,1] [,3] and [,2] making reference to the column 1 [Power], column 3 [slope] and column 2 [SFT.R.sq] corresponding to the table that we have obtained after running the function pickSoftThresold?

Fourth question, how can I know which networkType (signed, unsigned or signed hybrid) it corresponds for my data?

Fifth question, which information can we get of Scale Independence and Mean Connectivity plots?

Thank you,

Silvia

Answer 1

[ATpoint]

verbose simply defines how "talky" a function is so how much status messages it prints to screen. This has no effect on performance or output.

[Silvia Gonzalez]

Ok, and when you say status messages what does it means? What do you refer with status messages?

Thanks,

Silvia

[ATpoint]

No clue, never used the function myself. Just run it and see what kind of messages it prints to screen.

[Silvia Gonzalez]

Yes obviously, I have done with different values of verbose but I haven’t seen differences between the results. Anyway it’s ok.

Thanks for your help.

[Kevin Blighe]

Modifying the value of verbose should never change the results.

Answer 2

[Kevin Blighe]

Why at first we put 1:10 in a scale of one by one and then when we arrive at 12 we start to put the range two by two?

It is just testing different values for the soft thresholding power, with greater precision in the range 1:10. You can technically choose any values that you want.

Second, what it means the “verbose”? How can I know which verbose it’s the better for my data?

See the answer by ATpoint

Third question, when we are doing the Scale Independence plot we put the following code:

...In the function fitIndices we put [,1] [,3] and [,2] making reference to the column 1 [Power], column 3 [slope] and column 2 [SFT.R.sq] corresponding to the table that we have obtained after running the function pickSoftThresold?

Yes, and this plot helps you to choose the ideal soft thresholding value.

Fourth question, how can I know which networkType (signed, unsigned or signed hybrid) it corresponds for my data?

Please take a read of the answer from the WGCNA developer:

• https://support.bioconductor.org/p/84835/#84842

Fifth question, which information can we get of Scale Independence and Mean Connectivity plots?

For understanding 'connectivity' in a general sense in graph theory, I would perform a search in your search engine of choice. For understanding why we look at connectivity and scale independence in WGCNA, I would read this short [but good] answer:

• https://support.bioconductor.org/p/87024/#87034

[Silvia Gonzalez]

Ok, thank you for your attention