|Institution:||University of Pennsylvania|
|Keywords:||complex disease; copy number variation; exome sequencing; microarray; single nucleotide polymorphism; Bioinformatics; Genetics|
|Full text PDF:||http://repository.upenn.edu/edissertations/1099
In the human genome, DNA variants give rise to a variety of complex phenotypes. Ranging from single base mutations to copy number variations (CNVs), many of these variants are neutral in selection and disease etiology, making difficult the detection of true common or rare frequency disease-causing mutations. However, allele frequency comparisons in cases, controls, and families may reveal disease associations. Single nucleotide polymorphism (SNP) arrays and exome sequencing are popular assays for genome-wide variant identification. To limit bias between samples, uniform testing is crucial, including standardized platform versions and sample processing. Bases occupy single points while copy variants occupy segments. Bases are bi-allelic while copies are multi-allelic. One genome also encodes many different cell types. In this study, we investigate how CNV impacts different cell types, including heart, brain and blood cells, all of which serve as models of complex disease. Here, we describe ParseCNV, a systematic algorithm specifically developed as a part of this project to perform more accurate disease associations using SNP arrays or exome sequencing-generated CNV calls with quality tracking of variants, contributing to each significant overlap signal. Red flags of variant quality, genomic region, and overlap profile are assessed in a continuous score and shown to correlate over 90% with independent verification methods. We compared these data with our large internal cohort of 68,000 subjects, with carefully mapped CNVs, which gave a robust rare variant frequency in unaffected populations. In these investigations, we uncovered a number of loci in which CNVs are significantly enriched in non-coding RNA (ncRNA), Online Mendelian Inheritance in Man (OMIM), and genome-wide association study (GWAS) regions, impacting complex disease. By evaluating thoroughly the variant frequencies in pediatric individuals, we subsequently compared these frequencies in geriatric individuals to gain insight of these variants' impact on lifespan. Longevity-associated CNVs enriched in pediatric patients were found to aggregate in alternative splicing genes. Congenital heart disease is the most common birth defect and cause of infant mortality. When comparing congenital heart disease families, with cases and controls genotyped both on SNP arrays and exome sequencing, we uncovered significant and confident loci that provide insight into the molecular basis of disease. Neurodevelopmental disease affects the quality of life and cognitive potential of many children. In the neurodevelopmental and psychiatric diseases, CACNA, GRM, CNTN, and SLIT gene families show multiple significant signals impacting a large number of developmental and psychiatric disease traits, with the potential of informing therapeutic decision-making. Through new tool development and analysis of large disease cohorts genotyped on a variety of assays, I have uncovered an important biological role and disease impact of CNV in complex disease.