Integrating sequence conservation features and a domain-domain interaction network to detect disease-associated nsSNPs

Although several successful stories have been produced by genome-wide association (GWA) studies that are based on the common disease-common variant (CD-CV) hypothesis, recent studies have suggested a more general common disease-rare variant (CD-RV) hypothesis in the mapping of disease-associated genetic variants. Consequently, a number of statistical approaches have been proposed to detect associations between rare variants and inherited human diseases. Nevertheless, most of these methods require the selection of functional variants before applying statistical analysis, for the purpose of maximizing the power of statistical tests. To meet this requirement, we propose in this paper a filtration approach to detect genetic variants that are associated with a query disease of interest from the perspective of one-class novelty learning. Focusing on a typical type of genetic variants called nonsynonymous single nucleotide polymorphisms (nsSNPs), we propose to prioritize candidate nsSNPs relying on the integrated use of two sequence conservation features and a domain-domain interaction network. We resort to large-scale leave-one-out cross-validation experiments to assess the effectiveness of the proposed approach and show the power of this approach in detecting nsSNPs that are truly associated with the specific type of disease of interest.