Statistical Methods, Analyses and Applications for Next-Generation Sequencing Studies.

next-generation sequencin;Genetics;Health Sciences;Biostatistics

Lo, Yan YancyJohnson, Timothy D ;

University of Michigan

关键词: next-generation sequencin;    Genetics;    Health Sciences;    Biostatistics;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/116761/yancylo_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Current genetics studies rely heavily on next-generation sequencing (NGS) techniques. This dissertation addresses methodological developments and statistical strategies to efficiently and accurately analyze the large amounts of NGS data, thereby to understand the genetic contributions to diseases.In chapter 2, we evaluated the benefits of different variant calling strategies by performing a comparative analysis of calling methods on large-scale exonic sequencing datasets. We found that individual-based analyses identified the most high quality singletons, but had lower genotype accuracy at common variants than population-based and LD-aware analyses. Therefore, we recommend population-based analyses for high quality variant calls with few missing genotypes, complemented by individual-based analyses to obtain the most singleton variants.In chapters 3 and 4, we addressed the issue of overlapping read pairs in NGS studies arising from short fragments. In chapter 3, we proposed novel models to separately estimate machine and fragment errors of a NGS experiment from overlapping read pairs. Using a Markov chain Monte Carlo algorithm, our models suggested that machine and fragment errors were largely predicted by the reported quality scores of the overlapping bases and were uniform across individual samples from the same experiment. In chapter 4, we proposed an algorithm, RESCORE, to resolve the fragment dependence while retaining machine error estimates in overlapping reads. When compared to soft-clipping the overlapping regions, RESCORE increased the recalibrated base quality scores for the majority of overlapping bases, leading to a decrease in estimated false positive rate of novel variant discovery. In chapter 5, we presented an application of whole-genome sequencing for understanding the evolutionary history of uropathogenic Escherichia coli (UPEC). We sequenced 14 UPEC and 5 commensals at >190x, and found a deep split between UPEC and commensal E. coli. We observed high between-strain diversity, which suggests multiple origins of pathogenicity. We detected no selective advantage of virulence genes over other genomic regions. These results suggest that UPEC acquired uropathogenicity a long time ago and used it opportunistically to cause extraintestinal infections. In summary, this dissertation presented practical strategies for NGS studies that will contribute to further genetic advances.