Identifying SNP Interaction Patterns Using an Intensive Approach

• Activity Number: 134
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Hui-Yi Lin* and Dung-Tsa Chen and Po-Yu Huang and Yung-Hsin Liu and Chia-Ho Cheng and Jong Park
• Companies: Louisiana State University Health Sciences Center and Moffitt Cancer Center and National Chung-Hsing University and INC Research and Moffitt Cancer Center and Moffitt Cancer Center
• Keywords: interaction ; polymorphism ; SNP

Abstract

Clinical usage of the genome-wide association studies (GWAS) SNPs is limited. SNP-SNP interactions may be the key to overcome bottlenecks of genetic association studies. The related statistical methods are still under-developed. The objective of this study is to propose a new statistical approach for evaluating pairwise SNP-SNP interactions associated with a binary outcome. The most commonly used approach is the full interaction model (AA_Full), including two main effects and the interaction with an additive mode. This conventional approach is not sufficient. We proposed a new statistical approach, which is composed of 45 biologically meaningful interactions models. These patterns take SNP comparative allele, inheritance mode, and risk category grouping into consideration. Based on our simulation study, our approach had higher power than AA_Full approach and can overcome instable nature of SNP-SNP interaction patterns. For external validation, our approach allows flexibility to detect similar interaction patterns. We also applied our method to a large-scale prostate cancer study, some promising interactions were identified in both the discovery and validation sets.