Question

WGCNA soft threshold slope

0

Entering edit mode

kiddos17 • 0

@kiddos17-24245

Last seen 4.2 years ago

Hello All,

I was wondering what the value of the slope I am running wgcna with 150 samples/~4000 genes.

My r2 value plateus from 10 onwards with a value of 0.92+

At 10 the slope is -8.9 At 12 the slope is -5.5 At 14 the slope is -3.47 It then drops to -2 and then stays between -1 and -2.

I understand the slope should be negative. I've also read that it should be around -1. I don't understand if/why -1 is better than say -9.

But also if slope is changing so much does that mean the thresholding powers are not that stable

Thanks

TB

wgcna • 1.1k views

ADD COMMENT • link written 4.2 years ago by kiddos17 • 0

0

Entering edit mode

If you could post all the data to select the soft threshold it would be easier to understand what might be going on. Also, are your 150 samples homogeneous (with the same condition)?

ADD REPLY • link 4.2 years ago Lluís Revilla Sancho ▴ 760

score 0 · Answer 1 · 2020-09-30

Personally I would ignore the slope - IMO it's one of the less important statistic in the scale free topology analysis. With 140 samples, a power of 10 is reasonable for a signed network but seems too high for unsigned or signed hybrid network. If you're working with unsigned or signed hybrid networks, I would check the sample clustering and possibly PCA for large effects.