Statistical Methods and Analysis for Human Genetic Copy Number Variation and Homozygosity Mapping

Single nucleotide polymorphism (SNP) arrays are used primarily for genetic association studies, with data being analyzed in most cases one SNP at a time. Several other applications of SNP arrays, however, involve integration of data over multiple markers for a single individual. Two such applications of SNP arrays are studies of copy number variants (CNVs) and regions of homozygosity or identity by descent. Hidden Markov models are a common approach to both of these problems, but other methods have been used as well. In this dissertation I address several methodological issues related to these two types of analysis, and also apply the methods to several datasets. \ud The purpose of my studies in CNVs is to better detect and analyze CNVs. A major concern for all copy number variation (CNV) calling algorithms is their reliability and repeatability. I use family data as a verification standard to evaluate CNV calling strategies and methods. I make recommendations for how CNV calls can be used in genome-wide association studies. I then apply them to analyze CNVs in studies of psychiatric disorders and birth outcomes. Results from these studies have the potential for great public health significance, because they can lead to better understanding of the genetic etiology and eventually to better markers for disease screening and diagnosis. \ud Homozygosity mapping is a powerful method to map genes for rare recessive disorders. However, current methods are not ideal, especially when using high density SNP array data from consanguineous families. This study develops improved methods for homozygosity mapping using dense SNP data, and thus will improve the ability of geneticists to find genetic causes of rare recessive diseases. Many of these rare disorders are life-threatening; identification of the disease genes may help with early diagnosis and treatment.\u