I have been looking for some time to find an answer to the following question. I would be grateful for any help/advice or directions.

I work very frequently with large number of proteins and assess their importance in a biological network. Traditionally there are two methods used to visualise proteins in a network and withdraw conclusions purely on network architecture bases, before one looks at the top hits in details using databases. One is degree and the other is clustering coefficience.

Degree refers to a node being connected to edges. The higher the degree, the more nodes are connected to a given node. Traditionally people interpret nodes with high degree as being biologically important, so called hubs. Since hubs are so highly connected it means that a failure of a biological hub has catastrophic consequences since the functions of many proteins are related/dependent on hubs.

However what confuses me is the biological meaning of clustering coefficient of a node in a given network. I’m aware that clustering coefficience refers to the tendency of a nodes neighbours to connect to each other; however, I was unsure as to what it means biologically and whether that means a node with a high clustering coefficience is more “important” than a node with low clustering coefficience. This just seems unlikely to me because hub proteins, are considered important proteins; however, hubs usually have low clustering coefficience but that doesn’t mean they are less important since they are highly connected (high degree). Therefore to put it in plain english, why do people use clustering coeffience in biological networks and what biological information is it providing them?

So I would appreciate any explanation/advice and references you could provide that would help me understand the biological meaning/interpretation of clustering coefficience.


1 Answer 1


First of all, I find this an interesting question.

In short:

  • Clustering coefficience is widely used

  • It is widely accepted that hub genes are more essential than poorly connected genes

  • but research is not always interested in the hub genes / proteins / molecules (for example most metabolic pathways involve water)

Answering your question, why people would want to use this kind of method: If you would take a random model (pathway) most of the lines will be equally distributed between the nodes. This is not the case for most biological networks. Because of this, the global clustering coeffiecence can be used to see if a network could be biological. Clustering coeffiecence can also be used to find out about more specific nodes. For example if a protein has a relationship with two other proteins (binding, regulation, etc.), the two other proteins are also more likely to have a relationship with each other (similar to a social network).





  • $\begingroup$ Yes, the part about clustering coefficient of a specific node answers my question! I also though perhaps a node with a high clustering co-efficient shows a somewhat redundancy of that node in a sub-cluster of a network since if its connection with its immediate neighbours is terminated, that nodes neighbours would still remain connected. Would that be a correct interpretation too? Many thanks for the references! $\endgroup$ May 9, 2014 at 20:28
  • $\begingroup$ I think that depends on what the nodes represent. In the case of hub genes I don't think redundant is the correct word (as water is obviously necessary for metabolism). $\endgroup$
    – Traple
    May 12, 2014 at 10:34

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .