Efficient haplotyping for families

Hapi is a novel dynamic programming algorithm for haplotyping nuclear families that outperforms contemporary family-based haplotyping algorithms. Haplotypes are useful for mapping and identifying genes which cause and contribute to the etiology of human disease, and for analyzing the products of meiosis to locate recombinations, enabling the identification of recombination hotspots and gene conversions. They can also be used to study population history, including expansion, contraction, and migration patterns in humans and other species. Hapi's efficiency is a result of eliminating or ignoring states and state transitions that are unnecessary for computing haplotypes. When applied to a dataset containing 103 families, Hapi performs over 3.8-320 times faster than state-of-the-art algorithms. These efficiency gains are practically important as they enable Hapi to haplotype family datasets which current algorithms are either unable to handle or are impractical for because of time constraints. Hapi infers both minimum-recombinant and maximum likelihood haplotypes, and because it applies to related individuals, the haplotypes it infers are highly accurate over large genomic distances. Thesis Supervisor: David K. Gifford Title: Professor of Electrical Engineering and Computer Science Efficient Haplotyping for Families