C.J. Willer(1), L.J. Scott(1), H.M. Stringham(1), T.T. Valle(2), N.A. Rosenberg(3), R.N. Bergman(4), K.L. Mohlke(5), J. Tuomilehto(2,6), F.S. Collins(7), M. Boehnke(1)
(1) Dept of Biostatistics, University of Michigan, Ann Arbor, MI.; (2) Dept of Epidemiology and Health Promotion, National Public Health Institute, Helsinki, Finland; (3) Dept of Human Genetics & Life Science Institute, University of Michigan, Ann Arbor, MI; (4) Dept of Physiology and Biophysics, Keck School of Medicine, University of Southern California, Los Angeles, CA; (5) Dept of Genetics, University of North Carolina, Chapel Hill, NC; (6) Dept of Public Health, University of Helsinki, Helsinki, Finland and South Ostrobothnia Central Hospital, Seinjoki, Finland; (7) Genome Technology Branch, National Human Genome Research Institute, Bethesda, MD
We are carrying out a genomewide association scan of >300,000 SNPs on 2363 Finnish individuals using the HumanHap300 BeadChip from Illumina. Here we evaluate possible geographic stratification of SNP allele frequencies across 12 historical provinces of Finland. Our sample consists of type 2 diabetic individuals and non-diabetic controls, matched for province of birth. Each province is represented by 78 to 300 genotyped individuals. In our initial analysis of autosomal SNPs genotyped for 1743 individuals, we find an excess of significant allele frequency differences among the 12 provinces and between all pairs of provinces. Even the most similar provinces show a 9-fold excess of significant allele frequency differences at p <.0001. The pairs of provinces showing the strongest genetic differences are geographically distant (for example, Vaasa on the west coast and North Karelia in the east) and our results are in agreement with reported historical migration patterns and regional subisolates. These results suggest that careful case-control matching may be required to avoid false-positive results in genomewide association scans, even in populations often described as homogeneous.