An Overview of Haplotyping via Perfect Phylogeny: Theory, Algorithms and Programs

The next high-priority phase of human genomics will involve the development of a full Haplotype Map of the human genome. It will be used in large-scale screens of populations to associate specific haplotypes with specific complex genetic-influenced diseases. A key, perhaps bottleneck, problem is to computationally determine haplotype pairs from genotype data. An approach to this problem based on viewing it in the context of perfect phylogeny was introduced in 2002 along with an efficient solution. A slower (in worst case) variation of that method was later implemented. Two simpler methods for the perfect phylogeny approach that are also slower (in worst case) than the first algorithm were later developed. We have implemented and tested all three of these approaches in order to compare and explain the practical efficiencies of the three methods. In this talk I will introduce the haplotyping problem and the three algorithms for its solution that have been implemented. I will discuss two empirical observations: a strong phase-transition in the frequency of obtaining a unique solution as a function of the number of individuals in the input; and results of using the method to find nonoverlapping intervals where the haplotyping solution is highly reliable, as a function of the level of recombination in the data. Finally, I will discuss the biological basis for the size of these tests.