Centrality of cancer-related genes in human biological pathways: A graph analysis perspective

This study investigates standard and novel centrality models to identify the topological organization of cancer-related genes in biological pathways. We examined the linear relationship between the ratio of cancer-related genes and centrality rankings from different models. We also compared the cumulative distributions of centrality scores for cancer-related and non-cancer-related genes. Difference between the mean centrality scores of the two groups was tested in each pathway. The results show that when accounting for the directions of pathways and the importance of the interacting genes, the centrality of a gene correlates with the probability of cancer-relatedness. In particular, we show that the centrality measures we propose, namely Source-Sink PageRank and Source-Sink Katz, produce a distinction between the distribution of the two gene groups. Source-Sink PageRank shows the highest statistical power in differentiating between the means centrality values of two groups. The presented analysis provides a new perspective for understanding the topological organization of cancer-related genes.