Bioinformatic applications for virology research

Abstract

Viruses are one of the main classes of microscopic agents which cause infectious disease in humans. They have the potential to spread rapidly in a locality or even worldwide and infect a large proportion of the human population. Viruses such as influenza and HIV have affected millions people and result in the deaths of hundreds of thousands worldwide annually. To reduce disease mortality and risk of certain cancers in humans, early detection of viral infections is vital. As such, there has been continual development of virological tests to provide fast, accurate and cost-effective diagnosis. So far, these virological tests have proven to be essential for the management of viral infections and administration of treatment. However, the genetic arms race between viruses and host cells is never-ending. As host cells produce stronger immune responses to counteract the invading viruses, viruses evolve to enhance their ability to infect. As viruses evolve, they may become new variants or novel viruses with unpredictable virulence. Hence, the health threat that these new viruses will present cannot be overlooked. Early detection and continual biosurveillance of viruses, as well as understanding their evolution, are the solutions for preventing viral pandemics and controlling emerging infectious diseases. Over the years, a myriad of technology and methods have been developed to detect, obtain and analyze the genetic information of viruses to understand their virulence and evolution. This dissertation presents new computational tools and methods that improve upon existing approaches. Firstly, the thesis introduces a model that can predict amplification efficiency of random-tagged primers and uses it to as a basis to develop LOMA, an algorithm to design sensitive and efficient random-tagged primers. Subsequently, the thesis describes the design of microarrays and proposes novel analysis algorithms, PDA and EvolSTAR, for diagnostics and resequencing respectively. Lastly, the thesis presents RB-Finder, a fast distance-based sliding window algorithm that has accuracies comparable to phylogeny-based methods, to detect recombination breakpoints. With these innovations, the thesis aims to develop technologies and bioinformatics tools that have a greater impact on clinical decision-making.