Genome-wide association scan for type 2 diabetes in Finns. L.J. Scott¹, W.L. Duren¹, L.L. Bonnycastle², H.M. Stringham¹, A.U. Jackson¹, M.L. Erdos², P. Chines², N. Narisu², C.J. Willer¹, K.F. Doheny³, E.W. Pugh³, N.L. Riebowl², T.T. Valle⁴, J. Tuomilehto⁴, R.N. Bergman⁵, K.L. Mohlke⁶, F.S. Collins², M. Boehnke¹. 1) U Michigan, Ann Arbor, MI; 2) NHGRI, Bethesda, MD; 3) CIDR, J.H.U., Baltimore, MD; 4) National Public Health Institute, Helsinki, Finland; 5) U Southern California, Los Angeles, CA; 6) U North Carolina, Chapel Hill, NC.
   Our goal is to identify genes that increase the risk of type 2 diabetes (T2D). The FUSION study group is carrying out a genome-wide T2D association scan in 2357 first-stage samples of a two-stage study, which will ultimately include genotyping a total of ~5400 geographically matched Finnish samples. The Center for Inherited Disease Research (CIDR) has performed the sample genotyping using the Illumina HumanHap300 BeadChip. We report here an intermediate analysis of the first stage, based on of 885 T2D cases and 885 normal glucose tolerant controls and 308,231 polymorphic autosomal SNPs. In this sample, using a conservative genome-wide significance level of .05/302K = 1.6 x 10^-7, we have 12% or 87% power to detect an additive OR of 1.3 or 1.5 with a risk allele frequency of 0.3. The average duplicate error rate per genotype was 0.002% and the average SNP completeness was 99.6%. We flagged 5357 SNPs for future consideration based on deviation from Hardy-Weinberg equilibrium, elevated error rates, low completion rates or low minor allele frequency. We analyzed 302,874 SNPs for T2D association using dominant, recessive and additive models and adjusted the results to account for the 3 tests. Our most significant SNP had a p-value of 5.5 x 10^-6. We examined results around SNPs with multiple reported associations with T2D. Near PPARG Pro12Ala, TCF7L2 rs12255372, and KCNJ11 Glu23Lys, the most significant SNPs had p-values of .005, .005, and .03, and were ranked 1648^th, 1831^st, and 10321^st, respectively. These results suggest that other T2D susceptibility SNPs may be detected after genotyping the top 1-3% of SNPs in the full two-stage sample. To increase the power of the analysis, we will incorporate statistical weighting based on annotation information, stratify the sample by age-of-onset, and combine results across multiple T2D studies.