232

Bioinformatics of the Brain

FIGURE 9.3

a Erdos-Renyi Network, b Watts-Strogats Network.

FIGURE 9.4

a Barabasi-Albert Network 20 nodes, b Barabasi-Albert Network 40 nodes.

9.4

Modules and Hubs

A general property of complex networks is the existence of densely connected

subnetworks which are commonly called modules. This behavior can be ob-

served in various real networks such as social networks and the Web. Yet

another significant property of these networks is their hierarchical organiza-

tion such that a number of modules may be contained in a larger module.

Brain networks are no exception to these properties of complex networks;

they are small-world, scale-free networks containing modules in a hierarchical

structure.

9.4.1

Background

The problem of detecting such modules is commonly termed as clustering and

when the brain networks is modelled by a graph, the problem is referred to as

graph clustering which aims to find the densely connected regions of a brain

network. A hub in a complex brain network is a node that is highly connected