Dear Caitlin

Lets check our understanding of these measurements:

1) kTotal - connectivity of the each gene based on its r-values to all other genes in the whole network

2) kWithin - connectivity of the each gene within a single module based on its r-values to all other genes within the same module

3) and 4) kOut and kDiff mathematical derivatives from 1) and 2)

So, let say WGCNA identified 10 modules, but kWithin for "Module 2" is the largest and obviously larger than kTotal. This suggest "Module 2" to be a core of the network, or more important. Let say You perform functional annotation of the genes in this module and will find them to be enriched in amino acid analysis. And Your study was dedicated to metabolic changes under some conditions. So You can make conclusion that Your condition strongly affects AA metabolism, and via it - whole network.

In contrast, high kOut can suggest that total connectivity is much larger that connectivity within modules, and I would say - reflects sort of network's stability under tested conditions. So a set of vertices with high kOut can be targets for annotation as hubs that determined this stability. And elimination of them (knockout mutations, for example) can break the whole network.

Does it make sense?

Also, a comparison of these connectivity measures with kME would be great. I am trying to understand which measures are best for analyzing my data. Thanks!

Dear Caitlin

It is quite difficult to interpret this without a picture, but in general (as far as I understood) You can compare a level of significance of hub genes between different modules. Sometimes it can be very beneficial in finding the genes  to focus on (with high kDiff or kWithin for example) for functional analysis, for example.

Hi Albert

Thank you for your response. Do you know how to interpret kDiff or kOut? What do those measurements say about the significance of a gene in the whole network or within a module? Thanks again!

Dear Caitlin

Lets check our understanding of these measurements:

1) kTotal - connectivity of the each gene based on its r-values to all other genes in the whole network

2) kWithin - connectivity of the each gene within a single module based on its r-values to all other genes within the same module

3) and 4) kOut and kDiff mathematical derivatives from 1) and 2)

So, let say WGCNA identified 10 modules, but kWithin for "Module 2" is the largest and obviously larger than kTotal. This suggest "Module 2" to be a core of the network, or more important. Let say You perform functional annotation of the genes in this module and will find them to be enriched in amino acid analysis. And Your study was dedicated to metabolic changes under some conditions. So You can make conclusion that Your condition strongly affects AA metabolism, and via it - whole network.

In contrast, high kOut can suggest that total connectivity is much larger that connectivity within modules, and I would say - reflects sort of network's stability under tested conditions. So a set of vertices with high kOut can be targets for annotation as hubs that determined this stability. And elimination of them (knockout mutations, for example) can break the whole network.

Does it make sense?

Hi Albert,

Thank you so much! Your response was excellent and just what I was looking for. Considering each measurement generates slightly different top gene lists it helps to start thinking about connectivity measures as tools for different types analysis within the data. It seems obvious now, your comments make it more clear what we are getting out of the analysis for our data. I hope I am understanding this well. Thank you again!