We next inferred haplotypes using PHASE and the genotype data from SNP 1 to 35. Using the data from each of these SNPs, 6 haplotypes with a frequency of greater than 1% were observed (Figure 2). Three of the haplotypes (Haplotypes 1, 2 and 3) accounted for the vast majority of the observed haplotypes, with three other haplotypes occurring at frequencies between 1 and 2%. In total, these six haplotypes accounted for ~93% of the total observed haplotypes. In addition, there were twenty haplotypes that were observed only once. Visual inspection of each of these haplotypes which were formed using data from all 35 SNPs showed that almost all of these singularly observed haplotypes differed from the six major haplotypes by only one genotype. Since the haplotype information for these 516 subjects represent 18,060 individual genotypes, even if we assume that our genotyping accuracy rate is 99.9%, it is probable that a substantial portion of these haplotypes that were observed only once are actually artifacts secondary to genotyping errors. Therefore, we re-imputed haplotypes using the data from a set of
