PMCHAP: a new approach for haplotype inference using nuclear families with only one parent and multiple offspring

A new approach is suggested to perform joint haplotype reconstruction and haplotype frequency estimation in families consisting of multiple full sibs where only one parent is genotyped. It can also deal with the mixed data structures with 0 and 1 parent genotyped. The approach is based on the maximum likelihood principle using the expectation-maximization algorithm and thus combines pedigree information within families and linkage disequilibrium information between families with assumption of no recombination. In a simulation study of single nucleotide polymorphisms, the new algorithm termed PMCHAP (parent multiple children haplotyping) is shown to provide significantly better results than an approximation used so far, especially providing better estimates with increasing family sizes. It will be especially helpful for research in multiparous species like pigs, dogs, fish, and many lab animals where it is easy to collect families with multiple siblings.