Genome-wide association studies have been successful in identifying novel susceptibility loci across a range of complex diseases. However, the underlying mechanisms by which associated variants influence disease or quantitative phenotypes is often undefined and the variance that they explain is almost invariably small. Next Generation Sequencing (NGS) has allowed rarer genetic variants, which may potentially contribute substantially to phenotypic variance, to be genotyped and imputed in larger samples than before. The analysis of these variants may prove vital in understanding the missing heritability of complex disease. This has inspired the development of methodology to detect effects from multiple rare variants (termed collapsing methods), which single variant approaches may have limited power to detect. This thesis consists of several novel approaches to analysing regions of-rare genetic variation. Chapter 2 demonstrates an innovative approach to filtering variants prior to applying collapsing methods. In chapters 3 and 4 this approach is implemented in conjunction with collapsing methods to networks and pathways of genes respectively, identifying association signals which would not be identified using conventional filtering methods or by analysing genes individually. A similar method is applied to regions which map to protein domains in chapter 5, although these analyses provided limited evidence of association which was not driven by single variant effects. Analyses in chapter 6 u:vestigate the application of collapsing methods to DNA methylation data. Analysing the combined effect of variants provided evidence of association with methylation levels which were not identified using single variant approaches. Conventional applications of collapsing methods using individual gene regions and filtering by variant consequence have typically resulted in underpowered analyses. Results in this thesis demonstrate that collapsing variants using alternative definitions of a functional unit, filtering by predicted deleterious impact and analysing their effect on epigenetic phenotypes can identify novel association signals. Whilst collapsing methods remain a relatively new paradigm in statistical genetics, future endeavours which analyse increasingly larger samples of low coverage sequence data should benefit from using the approaches presented in this thesis to uncover the genetic architecture of complex disease.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:701970 |
| Date | January 2015 |
| Creators | Richardson, Thomas Golden |
| Publisher | University of Bristol |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0021 seconds