What is important graph parameters in PPI analysis?

Question

[Saeed Rahmani]

I want to analysis of PPI for Normal and Cancer cells, I use graph theory to this work. Which graph parameters important and suitable to this analysis?

Is Degree Centrality measure suitable? and this parameter show which property of PPI.

[Prakki Rama]

I worked on degree centrality before for my thesis (unpublished), so I am posting here to share my experience.

I used more than 20 PPI networks to find the essential proteins using degree centrality. Although some proteins in some networks had very high number of connections with other proteins, sometimes they were not found to be essential proteins in the network. At the end of it, I understood that to find an essential protein, sometimes other centrality measures such betweeness, subgraph play a significant role. So, choosing a centrality measure depends on the property of the network connections.

Answer 1

[Jean-Karim Heriche]

There are many different measures of centrality, all informing on different aspects of the graph. Have a look at this post for some simple explanations of commonly used centrality measures. Which measure you use depends on what features of your graph you're interested in.

[Saeed Rahmani]

I need a paper that describe biological term of graph parameter in biological, for example what is show with closeness centrality in a PPI network? and other parameter, closeness, betweenness, diameter and ...

[Jean-Karim Heriche]

The centrality measures relate to the structure of the graph. Their biological interpretation depends on the type of biological data and biological knowledge you can associate with the graph. This paper reviews and illustrates a lot of different centrality measures (but, despite the title, is rather light on the biology). The tutorial for the cytoscape plugin CentiScaPe gives some biological interpretation of the centrality measures implemented in the plugin.

Answer 2

[ivivek_ngs]

You can take a look at this thread and also look for the threads that pops out with your question on the right as similar posts. They might be helpful for the methods to assess PPI network using different modules of graph theory.

I think you should go to the 2 recent papers here and here to understand why the usage of centrality is important and what parameters are required to have the most enriched interaction network that is statistically viable and give much context driven biological meaning. Even there is a tool known as http://string-db.org/, you can go through its paper and several posts concerning the same is also available.