The estimation of haplotypes from SNP genotypes, commonly referred to as ‘phasing’, is a well studied problem in the literature. Existing approaches can be categorised according to the type of cohort each method is designed to phase, and the level of relatedness between the individuals in that cohort. Much of the recent literature is devoted to phasing nominally unrelated (or distantly related) individuals. Currently, the most accurate methods use hidden Markov models (HMMs) to model local haplotype sharing between individuals [1], [2], and take advantage of linkage disequilibrium (LD). Some of these methods can also handle mother-father-child trios and parent-child duos [2]–[6], for more complex pedigrees there are several general pedigree analysis software packages [7]–[10].
