PSB 2007 Tutorial
Technological advances in the genomic era has enabled extensive and affordable characterization of genomic variation and enhanced opportunity to identify regions associated with complex human diseases and traits. This 3 hour tutorial will focus on various statistical analysis and study design issues that arise from conducting Genome-Wide Association Scans (GWAS). We will highlight the differences between linkage and association, show how linkage disequilibrium can be exploited for association studies, and discuss the implications of population stratification and the dangers of spurious findings. We will explore the relative merits of family-based studies versus studies utilizing unrelated individuals, and corresponding implications for power. There are several online database resources that can be used to aid in the design and interpretation of these experiments, such as the HapMap, dbSNP, and the Pharmacogenetics Research Network Knowledge Base. We will consider the problems of heterogeneity and genetic or environmental interactions and demonstrate various lumping and splitting techniques that can reveal such effects. Finally, we will highlight some of the special statistical challenges of distinguishing signals from noise in the face of the huge number of multiple comparisons generated in GWAS, and demonstrate some of the newer methods and approaches to this problem.