PRIORITIZING DISEASE GENES BY INTEGRATING DOMAIN INTERACTIONS AND DISEASE MUTATIONS IN A PROTEIN-PROTEIN INTERACTION NETWORK

Abstract

Complex diseases such as cancer are involved in inter-relationship among several genes, with protein-protein interaction networks being extensively studied in attempts to reveal the relationship between genes and diseases. Although these studies have shown promising results for identifying disease genes, it is not systemically studied that a protein functions differently depending on its interaction partners in the network since a protein can have multiple functions. In this study, domains are considered as functional units of proteins and we investigate how disease-related mutations in domains can be used to identify other disease genes in a domain-domain interaction network. We subsequently propose a computational method to predict disease genes based on the following two assumptions. The first assumption is that proteins closely interacting with known disease proteins in a protein interaction network are likely to be involved in the same disease. Second, although two proteins are in the same distance from known disease genes in a protein interaction network, the protein interacting with known disease genes through a domain with mutation is more likely to be related to the disease than other proteins that interact through domains with no mutation. As a result, when the proposed approach is applied to five diseases, it highly ranks disease-related genes compared to a model using only a protein interaction data set.