1 results
10 - Bayesian Model Averaging for Genetic Association Studies
-
- By Christine Peterson, Rice University, Michael Swartz, The University of Texas, Sanjay Shete, The University of Texas, Marina Vannucci, Rice University
- Edited by Kim-Anh Do, Zhaohui Steve Qin, Emory University, Atlanta, Marina Vannucci, Rice University, Houston
-
- Book:
- Advances in Statistical Bioinformatics
- Published online:
- 05 June 2013
- Print publication:
- 10 June 2013, pp 208-223
-
- Chapter
- Export citation
-
Summary
Genetic Association Studies
Genetic association studies seek to identify genetic variants associated with a particular disease or phenotype. Typically, the phenotype of interest will be a complex trait that is not determined by a single recessive, additive, or dominant allele (Lander and Schork, 1994). Although there has been a shift in recent years toward genome-wide association studies(GWAS) that agnostically test all single-nucleotide polymorphisms (SNPs) for association with a disease, many genetic association studies still focus on candidate genes that have functions known to be related to the phenotype. Although it is scientifically preferable to include as many SNPs as possible, the number of SNPs in a study is often limited by practical constraints such as cost. The statistical methods discussed in this chapter are primarily applicable to studies using several hundred SNPs. For a discussion of Bayesian approaches for full-scale GWAS, which may include up to 1 million SNPs, see Chapters 9 and 11.
A variety of study designs for genetic association studies have been proposed; for a comparison, see Cordell and Clayton (2005). The most common design for binary outcomes is the population-based case-control study. Family-based designs, which examine patterns of inheritance in related individuals, are also useful, particularly when the variant of interest is rare in the overall population. Data for family studies may come from extended pedigrees or from triads of affected offspring and their parents. In the case-parent triad design, disease-associated SNPs are identified using the transmission disequilibrium test (TDT), which is based on the idea that affected children will have nonrandom inheritance of causal genetic variants from their parents.