Figure 1

The imputation framework (left) and an example illustrating the imputation procedure (right). In the example, individuals with grey color have single-nucleotide polymorphism (SNP) chip data from genome-wide association studies (GWAS), and individuals with black color have both chip data and sequence data. The haplotypes in this segment are labelled using different colors and they are inferred based on GWAS data. Notice that both haplotypes of individual 949 and one haplotype of individual 957 can be recovered based on the information of their children (the missed haplotype is illustrated using a thin black bar). However, only one haplotype can be recovered for 957 because he only has one child. The two variants are from sequence data (1 and 2 are alleles, and 0 is missing). For the first variant, because member 974 is homozygous genotype (1, 1), the alleles on its two haplotypes (pink and dark blue) can be assigned. Subsequently, the alleles on the light blue haplotype of member 940, the yellow haplotype of member 956, and the green haplotype of member 939 can be resolved (all three have sequence data). For all the other members, their alleles can be imputed based on the color of their haplotypes. However, haplotype light green (in members 949, 959, and 960) cannot be imputed because it has not occurred in any sequenced individual, thus showing missing one allele. For the second variant, our algorithm will identify a conflict because member 974 assigns allele 2 to the pink haplotype, and member 939 assigns allele 1 to the pink haplotype.