For GAD1, two blocks were defined consisting of 2 and 5 markers, respectively. However, LD between these two blocks is strong and yields a highly correlated haplotype distribution for the common haplotypes. The most common haplotype in block 1 (frequency 42.8%) occurs with the most common haplotype in block 2 (43.1%); the second and third most common haplotypes in block 1 are similarly related to the corresponding haplotypes in block 2. The one-to-one correspondence of the three common haplotypes results in a high correlation of 0.84 between the two blocks. Therefore, we collapsed the two blocks together, forming a long haplotype block with seven markers. For GAD2, LD was extremely high among 9 out of 11 genotyped markers, with a mean pair-wise D′ of 0.983 (D′ estimates for adjacent SNPs were greater than 0.95 for 97% of the comparisons). We used Tagger with r2 of 0.8 and MAF of 0.2 (de Bakker et al., 2005) to select the minimum numbers of markers to capture maximum haplotype variation within blocks for the GAD1 and GAD2 genes. Consequently, four (rs2241165, rs3828275, rs2058725, rs701492) and two (rs2839677, rs7900976) tagging SNPs were used for haplotype analyses in GAD1 and GAD2, respectively.
