WGCNA: Problem with selecting soft threshold?

Question: WGCNA: Problem with selecting soft threshold?

11 weeks ago by

Hello I have a problem to select the power to use I 14 sample of different time points I removed the lowely expressed genes and used the log normalized count for only diff expressed genes with is 2065 genes

powers = c(c(1:10), seq(from = 12, to=45, by=2))

sft = pickSoftThreshold(matt, powerVector = powers, verbose = 5, dataIsExpr = T)

this what I get

I picked power 42 is there anything wrong?

rna-seq soft thresholding wgcna • 222 views

ADD COMMENT • link •

modified 11 weeks ago by andres.firrincieli • 1.1k • written 11 weeks ago by shaimaa.m.hassan.18 • 0

11 weeks ago by

andres.firrincieli • 1.1k

andres.firrincieli • 1.1k wrote:

... and used the log normalized count for only diff expressed genes with is 2065 genes

From WGCNA faq:

First, the user should ensure that variables (probesets, genes etc.) have not been filtered by differential expression with respect to a sample trait. See item 2 above for details about beneficial and detrimental filtering genes or probesets.

Filtering for low expressed genes is fine.

14 sample

14 samples is a little bit to low for WGCNA, however: WGCNA sample size minimum: why?

I picked power 42 is there anything wrong?

42 is too high. Remove only the low expressed genes and repeat the analysis. Because of the low sample size things will not get better but, who knows...

ADD COMMENT • link written 11 weeks ago by andres.firrincieli • 1.1k

Hello Thank you !!! I now didnt removed the nonsig differentially expressed genes I only removed lowely expressed genes and I rerun it it is much better but it never reach 0.9 it reach 0.85 can I use it now ![enter image description here][1]

pickSoftThreshold: calculating connectivity for given powers... ..working on genes 1 through 5977 of 7485 ..working on genes 5978 through 7485 of 7485 ""Power SFT.R.sq slope truncated.R.sq mean.k. median.k. max.k. ""

0.847 2.120 0.950 3070 3130.0 4240

3 2 0.525 0.427 0.767 1750 1710.0 3060

4 3 0.136 -0.118 0.431 1150 1040.0 2420

5 3 0.136 -0.118 0.431 1150 1040.0 2420

6 4 0.715 -0.410 0.750 824 685.0 2010

7 5 0.819 -0.569 0.824 621 471.0 1720

8 5 0.819 -0.569 0.824 621 471.0 1720

9 6 0.853 -0.686 0.851 485 340.0 1500

10 7 0.868 -0.768 0.860 390 250.0 1330

11 7 0.868 -0.768 0.860 390 250.0 1330

12 8 0.866 -0.834 0.862 321 189.0 1190

13 9 0.862 -0.884 0.856 268 146.0 1070

14 9 0.862 -0.884 0.856 268 146.0 1070

15 10 0.872 -0.919 0.869 228 114.0 972

16 11 0.881 -0.949 0.880 196 90.7 889

17 13 0.863 -1.000 0.868 148 59.3 754

18 15 0.859 -1.040 0.873 116 39.9 649 ""

ADD REPLY • link modified 11 weeks ago • written 11 weeks ago by shaimaa.m.hassan.18 • 0

Glad it works now. From WGCNA faq:

If the scale-free topology fit index fails to reach values above 0.8 for reasonable powers (less than 15 for unsigned or signed hybrid networks, and less than 30 for signed networks) and the mean connectivity remains relatively high (in the hundreds or above), chances are that the data exhibit a strong driver that makes a subset of the samples globally different from the rest.

In your case I would pick 11

ADD REPLY • link written 11 weeks ago by andres.firrincieli • 1.1k

Similar posts • Search »