UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Efficient assembly of large genomes Jackman, Shaun Dunn


Genome sequence assembly presents a fascinating and frequently-changing challenge. As DNA sequencing technologies evolve, the bioinformatics methods used to assemble sequencing data must evolve along with it. Sequencing technology has evolved from slab gel sequencing, to capillary sequencing, to short read sequencing by synthesis, to long-read and linked-read single-molecule sequencing. Each evolutionary jump in sequencing technology required developing new bioinformatic tools to address the unique characteristics of its sequencing data. This work reports the development of efficient methods to assemble short-read and linked-read sequencing data, named ABySS 2.0 and Tigmint. ABySS 2.0 reduces the memory requirements of short-read genome sequencing assembly by ten fold compared to ABySS 1.0. It does so by using a Bloom filter probabilistic data structure to represent a de Bruijn graph. Tigmint uses linked reads to identify large-scale errors in a genome sequence assembly. Correcting assembly errors using Tigmint before scaffolding improves both the contiguity and correctness of a human genome assembly compared to scaffolding without correction. I have also applied these methods to assemble the 12 gigabase genome of western redcedar (Thuja plicata), which is four times the size of the human genome. Although numerous mitochondrial genomes of angiosperm are available, few mitochondria of gymnosperms have been sequenced. I assembled the plastid and mitochondrial genomes of white spruce (Picea glauca) using whole genome short read sequencing. I assembled the mitochondrial genome of Sitka spruce (Picea sitchensis) using whole genome long read sequencing, the largest complete genome assembly of a gymnosperm mitochondrion. The mitochondrial genomes of both species include a remarkable number of trans-spliced genes. I have developed two additional tools, UniqTag and ORCA. UniqTag assigns unique and stable gene identifiers to genes based on their sequence content. This gene labeling system addresses the inconvenience of gene identifiers changing between versions of a genome assembly. ORCA is a comprehensive bioinformatics computing environment, which includes hundreds of bioinformatics tools in a single easily-installed Docker image, and is useful for education and research. The assembly of linked read and long read sequencing of large molecules of DNA have yielded substantial improvements in the quality of genome assembly projects.

Item Media

Item Citations and Data


Attribution-NonCommercial-NoDerivatives 4.0 International