Methods for resolving genotype and haplotype ambiguity in human genetic data.

The identification and characterization of genes involved in human disease, especially complex diseases like cancer and diabetes, requires genetic data of high fidelity. For many diseases, particularly those of late onset, family material is limited and often inadequate for ensuring accuracy and completeness of genetic information. Common sources of ambiguity include errors and mutations in the genotypes, the inability to directly observe the maternal and paternal origins of the constituent alleles in each genotype, in other words, the haplotypes, and deceased or otherwise unavailable pedigree members. Three manuscripts form the basis of this dissertation. Each describes methods for genotyping error and mutation detection and haplotype construction when pedigree information is limited or unavailable, for example, in sib-pair or case-control studies. In the first manuscript, I present a hidden Markov method for detecting genotyping errors and mutations for a sib pair in the context of multipoint mapping. The method is designed for use with sib-pair data when parental genotypes are unavailable. Through Monte Carlo simulation, I explore the accuracy of my error-detection method and examine the impact of genotyping errors and error detection and correction on multipoint linkage information. In the second manuscript, I derive analytic formulae and simulate genetic data to estimate the expected rate at which genotyping errors appear as inheritance inconsistencies in nuclear family data. Numerical results from these investigations suggest that at least a quarter of genotyping errors are still undetectable on the basis of inheritance inconsistencies under even favorable circumstances, underscoring the importance of identifying errors that are consistent with Mendel's laws, for example, through multipoint analysis. In the third manuscript, I describe and evaluate an application of somatic cell hybrids for haplotype construction. The approach permits direct observation of individual haplotypes, thereby eliminating the need to collect and genotype DNA from family members for haplotype-based analyses. I present experimental data and evaluate the theoretical efficiency and cost effectiveness of using haplotypes instead of conventional genotypes in the context of haplotype frequency estimation. Each of these manuscripts addresses practical issues related to error detection and haplotype construction for gene mapping studies with incomplete genetic data.PhDBiological SciencesBiostatisticsGeneticsUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/125145/2/3016835.pd