a statistically more robust method for complex phenotypes (Risch and Merikangas, 1996). Association studies test for genomic variants with statistically different frequencies in affected versus unaffected persons. Association analyses can be conducted within families (in which case one is testing for linkage and association (Martin et al., 2000), or, more commonly now, across families (Hirschhorn and Daly, 2005; Manolio, 2010; McCarthy et al., 2008). Genetic association is the method used in both candidate gene studies and in GWAS. Candidate gene studies involve testing for association with genes hypothesized to be involved in a phenotype (e.g., alcohol metabolizing genes for alcohol use disorders) or those previously connected to a phenotype (e.g., in previously published reports or animal studies). In contrast, GWAS take an atheoretical approach and systematically search across the genome for significant associations. Thus, GWAS combine the unbiased nature of a linkage study with the statistical power of an association study. The genome-wide search has been aided by the identification and cataloging of variants, in particular single nucleotide polymorphisms (SNPs), across the genome [see the International Hapmap Project (Barrett et al., 2005) and the 1000 Genomes Project (http://www.1000genomes.org/) for details]. A SNP represents a single base pair change in the
