Predicting potential disease-related genes using the network topological features

To help the biomedical scientist pre-confirm the disease-related genes, we considered these gene as a whole research set and analyzed the topological features of their interaction network. Two strategies had been proposed to construct the disease-related gene network from the OMIM database. Using these two constructed sets, we trained two support vector machine prediction models, the accuracy of which are 75.09% and 83.63%. As a result, we gained 27 and 2873 potential disease-related genes respectively. The intersection of the two predicted sets contains 19 genes. In addition, gene locuses with high appearance frequency were listed for further research.