WGCNA analysis on transcriptomic data

Question

[imaparna27]

Hi,

I am trying to perform a WGCNA analysis for cancer transcriptomic data. I have taken value of soft-threshold(β) based on "power estimate" result returned by picksoftthreshold function of WGCNA. I have around 18 modules, however for downstream analysis I have taken the largest module based on the highest positive correlation value (0.77) in my analysis. However, I have another module having a significant number of genes with a very high negative correlation value (-0.85). Which module is better for downstream analysis? Also, what should be other criteria for module selection?

[andres.firrincieli]

can you be more specific? Are you talkning about the correlation values between genes of the same module or the correlation values you get from the module-trait relationship analysis?

[imaparna27]

correlation values for module-trait relationship analysis

[andres.firrincieli]

Which module is better for downstream analysis?

They are both important

[imaparna27]

So, how both can be used in downstream analysis? For instance I would consider intra-modular hub genes if I had to select only one module. But if i consider both modules seperately, how interpretation can be given if one suggests positive correlation with traits while other is showing negative correlation?

[andres.firrincieli]

You should always focus on the hub genes, no matter how many modules you pick for downstream analysis.

But if i consider both modules seperately, how interpretation can be given if one suggests positive correlation with traits while other is showing negative correlation?

You know the experiment and you know the biology; the interpretation of the results is up to you but technically speaking both modules are important. This is like asking "Should I consider for downstream analysis the up- or the down-regulated genes?"