METHODS FOR HIGH-THROUGHPUT COMPARATIVE GENOMICS AND DISTRIBUTED SEQUENCE ANALYSIS

Abstract

High-throughput sequencing has accelerated applications of genomics throughout the world. The increased production and decentralization of sequencing has also created bottlenecks in computational analysis. In this dissertation, I provide novel computational methods to improve analysis throughput in three areas: whole genome multiple alignment, pan-genome annotation, and bioinformatics workflows.

To aid in the study of populations, tools are needed that can quickly compare multiple genome sequences, millions of nucleotides in length. I present a new multiple alignment tool for whole genomes, named Mugsy, that implements a novel method for identifying syntenic regions. Mugsy is computationally efficient, does not require a reference genome, and is robust in identifying a rich complement of genetic variation including duplications, rearrangements, and large-scale gain and loss of sequence in mixtures of draft and completed genome data. Mugsy is evaluated on the alignment of several dozen bacterial chromosomes on a single computer and was the fastest program evaluated for the alignment of assembled human chromosome sequences from four individuals. A distributed version of the algorithm is also described and provides increased processing throughput using multiple CPUs.

Numerous individual genomes are sequenced to study diversity, evolution and classify pan-genomes. Pan-genome annotations contain inconsistencies and errors that hinder comparative analysis, even within a single species. I introduce a new tool, Mugsy-Annotator, that identifies orthologs and anomalous gene structure across a pan-genome using whole genome multiple alignments. Identified anomalies include inconsistently located translation initiation sites and disrupted genes due to draft genome sequencing or pseudogenes. An evaluation of pan-genomes indicates that such anomalies are common and alternative annotations suggested by the tool can improve annotation consistency and quality.

Finally, I describe the Cloud Virtual Resource, CloVR, a desktop application for automated sequence analysis that improves usability and accessibility of bioinformatics software and cloud computing resources. CloVR is installed on a personal computer as a virtual machine and requires minimal installation, addressing challenges in deploying bioinformatics workflows. CloVR also seamlessly accesses remote cloud computing resources for improved processing throughput. In a case study, I demonstrate the portability and scalability of CloVR and evaluate the costs and resources for microbial sequence analysis.