Clustering analysis

Looking for communities in a network is a nice strategy for reducing network complexity and extracting functional modules (e.g. protein complexes) that reflect the biology of the network. There are several terms that are commonly used when talking about clustering analysis (Figure 30):

Some concepts in network community analysis

Figure 30 Some concepts in network community analysis.  (Network communities figure from Wikimedia commons by j_ham3 and used under Creative Commons Attribution-Share Alike 3.0 Unported license. Haemoglobin 3D structure from PDBe and complex diagram from IntAct).

Community / Cluster 

A general, catch-all term that can be defined as a group of nodes that are more connected within themselves than with the rest of the network. The precise definition for a community will depend on the method or algorithm used to define it. When talking about PPINs, communities fall into two categories: functional modules and protein complexes.


In biology, modules are exchangeable functional units in which the nodes (proteins) do not have to be interacting in the same time or space. The most important characteristic of a module is that its intrinsic functional properties do not change when it is placed in a different context.


A complex is a group of proteins that interact with each other at the same time and in the same space, forming relatively stable multi-protein machinery. You can use the Complex Portal to explore known macromolecular complexes in a number of model organisms.


A subset of nodes in which every node is connected with every other member of the clique. A maximal clique is a clique that cannot be extended by adding an additional node not previously included in the clique. There are several different types of cliques and they can be used as the basis of algorithms that use topological criteria to look for communities.


Motifs are statistically over-represented sub-graphs in a network. They correspond with a pattern of connections that generates a characteristic dynamical response (e.g. a negative feedback loop). They are less important for the type of networks this tutorial is devoted to, but are quite useful in directed networks.

When exploring a PPIN for clusters, the goal is often to find functional modules or protein complexes that execute defined biological functions. There are many different methods that can help us find clusters and we will briefly introduce some of them in this section.