Bridging centrality: graph mining from element level to group level
Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining
| Hi-index | 0.00 |
The systematic analysis of protein-protein interactions is one of the most fundamental challenges to understand cellular organizations, processes and functions. The interaction between two proteins provides significant insight into their functional association. Recent high-throughput experiments have determined protein-protein interactions in the genome scale, called interactome. A wide range of graph theoretic, computational approaches have been presented to characterize protein functions from the interactome networks. In this work, we quantitatively analyze topological features of interactome networks. Our topological model is based on hierarchical modularity in the scale-free nature. First, we use connectivity and betweenness centrality to measure the likelihood of bridging two clusters for each node and edge. Next, we propose an efficient algorithm to detect clusters by collapsing the bridging nodes and edges. To assess the measurement of bridges, we compute the clustering coefficients of the networks which are built from successive deletion of bridging nodes. The alteration pattern of the clustering coefficients approximates the amount of bridging nodes in a network. We also investigate the biological importance of bridging nodes based on protein lethality information. The modularization results show that our approach accurately identifies functional modules in the interactome network of S. cerevisiae. We finally apply our approach to predicting biological functions of uncharacterized proteins in S. cerevisiae.