Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
975710 | Physica A: Statistical Mechanics and its Applications | 2014 | 12 Pages |
•We firstly take essential genes as reference to analyze human genes.•Nonessential disease genes are topologically more important than other genes.•Disease genes are not in the periphery but closer to the center than other genes.•The influence of disease genes on essential genes is weaker than other genes.•The new topological features are beneficial for disease genes prediction.
The topological features of disease genes and non-disease genes were widely utilized in disease genes prediction. However, previous studies neglected to exploit essential genes to distinguish disease genes and non-disease genes. Therefore, this paper firstly takes essential genes as reference to analyze the topological properties of human genes with protein–protein interaction network. Empirical results demonstrate that nonessential disease genes are topologically more important and closer to the center of the network than other genes (unknown genes, which are deemed as non-disease genes in disease genes prediction). Although disease genes are closer to essential genes, we find that the influence of disease genes on essential genes is similar with other genes, or even weaker. Further, we generate new topological features according to our findings and validate the effectiveness of combining the additional features for detecting disease genes. In addition, we find that the k-shell index (ks) of protein–protein network follows a power law distribution, and the function of the proteins with the largest ks may deserve further research.