Computational methods addressing genetic variation in next-generation sequencing data

Computational genomics involves the development and application of computational methods for whole-genome-scale datasets to gain biological insight into the composition and function of genomes, including how genetic variation mediates molecular phenotypes and disease. New biotechnologies such as next-generation sequencing generate genomic data on a massive scale and have transformed the field thanks to simultaneous advances in the analysis toolkit. In this thesis, I present three computational methods that use next-generation sequencing data, each of which addresses the genetic variations within and between human individuals in a different way. First, Samovar is a software tool for performing single-sample mosaic single-nucleotide variant calling on whole genome sequencing linked read data. Using haplotype assembly of heterozygous germline variants, uniquely made possible by linked reads, Samovar identifies variations in different cells that make up a bulk sequencing sample. We apply it to 13 cancer samples in collaboration with researchers at Nationwide Childrens Hospital. Second, scHLAcount is a software pipeline that computes allele-specific molecule counts for the HLA genes from single-cell gene expression data. We use a personalized reference genome based on the individual's genotypes to reveal allele-specific and cell type-specific gene expression patterns. Even given technology-specific biases of single-cell gene expression data, we can resolve allele-specific expression for these genes since the alleles are often quite different between the two haplotypes of an individual. Third, Vargas implements an optimal algorithm for edit distance alignment of sequencing reads to variant graph or linear reference genomes. We use these alignments to assess and improve the accuracy of several popular read alignment algorithms that use suboptimal (heuristic) algorithms. These methodological innovations enhance the capability of biotechnologies such as linked reads, single-cell gene expression, whole-genome sequencing, and RNA sequencing to reveal biological insights