A Computational Method for Identification of Functional SNPs in Human Noncoding Genome Regions based on Multi-feature Mining

<p>Single Nucleotide Polymorphism (SNP) is the variant on a single nucleotide in genome. Functional SNP, as one of most important molecular markers disease research, has been widely used various research fields, such tumor pathogenesis, diagnosis and treatment, prognostic evaluation, drug development, etc. The number functional SNPs noncoding genome regions much more than that coding regions, their detection difficult. In this work, multi-feature mining based computational method proposed to predict human genomes. We first analyzed sequence properties, evolutionary conservation properties epigenetic modification signal sample SNPs. Statistical methods together with multiple annotation data from genomes epigenetics were mine high-dimensional discriminative features subsequently. particular, allele-specific designed distinguish function close locations. random forest was conduct feature dimension reduction classification. 10-fold cross-validation result showed Area Under Receiver Operating Characteristic Curve (AUC) our improved by 16.9% 43.4% over existing GWAVA CADD, respectively, illustrating can help netural near locations.</p> <p> </p>