Allele frequency heterogeneity in the collaborative study on the genetics of alcoholism (COGA)

It is well known that admixture of populations having different marker allele frequencies can increase type I error in genome scans with ungenotyped parents, and can cause spurious association in case/control association studies of candidate genes. The Collaborative Study on the Genetics of Alcoholism has ascertained multiplex case families through alcoholic probands. Using markers from a genome screen, we compared allele frequencies among racial groups and found significant and extensive heterogeneity in the COGA data. Differences were also observed at candidate loci commonly studied for association with psychiatric disorders. A total of 609 genotyped founders from COGA case families have identified a racial group to which they belong. While these are broad groupings and diversity of background may be expected within as well as between groups, comparisons were made between the two most represented groups: Caucasian (N=506) and African-American (N=60), using Fisher's exact test. Comparisons of DRD2 and ADH3 were both significant at the 0.001 level; analyses of additional loci are ongoing. These findings confirm the need for caution when using linkage or association methods that depend on assumptions of homogeneity of allele frequencies. Evaluating allele frequency variations throughout the genome in these data will permit more informed genetic analyses.