How To Show A Ppi Subnetwork/Cluster Is Not Obtained By Random Chance?

Question

[Diwan]

Hi All,

I have created a large PPI sub-network using some genes associated with a disease. From this sub-network, I created network clusters using cytoscape clustering tool. Now I want to show that the network clusters are significant and cannot be obtained by random chance.

It will be helpful if you can point me towards any tool (or) suggest a way to accomplish this.

Thanks Diwan

Answer 1

[Sudeep]

I am not aware of any Cytoscape plugin that will serve your purpose, but you can go through this R igraph community detection tutorial, especially the last part "Testing the significance of a community" or this paper and the author has released his codes here.

[Diwan]

Thanks, Sudeep. Igraph looks nice. I will explore it more.

Answer 2

[B. Arman Aksoy]

For that, people usually show that their network of interest has a significantly deviant property compared to what they will get if they were to sample the whole network randomly by multiple times. As an example, you can take a look at Figure 2b of this paper, where they show this via the modularity property:

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0008918

[Diwan]

Thanks, Arman. This paper will perfectly fits my need.