Last week we hosted the visit of G.Scardoni, the author of Centiscape, a Cytoscape plugin to calculate different measures for Node Centralities in a network.
I recommend you to read the supplementary material 1 of his paper (it’s a pdf), because it has a good description of many measures of node centralities and their possible explication in a biological context.
A node centrality is a parameter that, given a node’s position and interactions in a network, determine its importance. To understand it better, think that one of the main purposes of centralities for biology is to identify which genes are more important in a biological process: which are in a bottleneck position, which are required for having a proper function and which ones are only redundant.
The simplest measure of node centrality is the Degree, which is the number of connections of a node. It seems logic to think that genes with an higher degree (higher number of interactions) should be more important than the others, because a loss of function there will affect more interactions. However, after reading at the Centiscape plugin I realized that there are a lot of measures for node centralities, including closeness, betweenness, stress, centroid, etc. The degree is not the best parameter to identify genes in bottleneck positions, for which we should use betweenness or stress instead.
Just to make this post round, I have opened a discussion on biostar about which measures of Node Centrality can be applied to biological networks. Let’s see what comes out from that discussion, and if there are other centralities I do not know yet :-).