| In order to benefit human health, advances in the genomics area must be translated into improvements in clinical care. A better understanding of the underlying biological and environmental factors that influence disease ("bench") will lead to the development of practical applications to directly benefit the outcome of patient care ("bedside"). To achieve this goal of translational research, the collaboration between practitioners and researchers from both (clinical and life sciences) domains is very important. A key component of translational research is the management, integration and analysis of large quantities of both clinical and high throughput genomic data. This dissertation focuses on research issues involved in making use of data from both of these areas.; In the area of high throughput genomic data, this dissertation discusses several issues concerning the integration of biological domain knowledge, such as pathway information, to supplement statistical and data-mining algorithms to explore the etiology of complex diseases. As an example, an in-depth association analysis of the complement pathway single nucleotide polymorphisms (SNPs) with age related macular degeneration (AMD) is provided. This dissertation also describes Pathway/SNP, a software tool that allows an exploratory approach to integrative association analysis.; Since most current genomic association studies use commercial genotyping platforms, an important area of current interest relates to understanding the characteristics and limitations of such "standardized" tools. This dissertation describes findings concerning two issues related to the scope of currently available commercial platforms: (1) the ability to identify copy number polymorphisms in tumors and (2) genomic coverage in different HapMap population samples.; In the area of clinical data, this dissertation discusses issues related to the use of Entity-Attribute-Value (EAV) database modeling, which is widely used in clinical data repositories. This dissertation discusses issues concerning the pivoting (transforming) of EAV data into one-column-per-parameter format before it can be used by a variety of analytical programs. It then broadly synthesizes the goals of EAV modeling, the situations where it is a useful alternative to conventional database modeling, and describes the fine points of its implementation in production systems. |
