Centrality in Biological Networks

Abstract

Centrality analysis has become an important part of biological network studies, notably that of protein-protein interaction networks. It has long been known that the importance of a protein is determined by its connections and relationships to other proteins. In the current work we look into centrality in other kinds of networks as well, notably those based on gene expression data and drug effects on cancer cell lines. The purpose of this project is to show that centrality is useful for the analysis of several different kinds of biological networks. Firstly, we show that the most central genes in the p53 protein interaction network are also the most relevant regarding the network’s ability to suppress tumors. Secondly, we look into different types of breast cancer and demonstrate that central genes are among the best discriminators between different classes of data. It is also interesting to see many of the p53 pathway elements coming up among the top central genes. Finally, we apply centrality on cancer treatment data and show how it can be used to identify good drug candidates for different kinds of cancer.