Computational approaches to understanding infectious disease
Infectious diseases derive from organisms such as viruses, bacteria, fungi and parasites that can be passed from person to person, transmitted via bites from insects or animals, or acquired through ingestion of contaminated food or water or environmental exposure. Infectious diseases cause roughly 20% of annual deaths worldwide, including many children under the age of five. In developing countries, these diseases remain a major public health problem. They can also cause societal and economic burdens through life-long disability. We need a better understanding of these diseases with a view towards the goals of prevention and cure. The advent of whole-genome transcriptional profiling technology and powerful computational resources has made it possible to study infectious diseases on a genome-wide scale. Such studies can lead to improvements in diagnostic tools as well as preventive measures such as vaccines. The work of this thesis focuses on a number of projects with the common thread of developing and applying of computational methods to extract biological information from high-throughput transcriptional data related to infectious diseases. These include (1) the identification of gene signatures related to B-cell proliferation that predict an influenza vaccine-induced antibody response; (2) study of the physiological state of the Plasmodium falciparum malaria parasite when sequestered in human tissue; (3) identifying the similarity and differences of the response to five anti-viral vaccines. To achieve the scientific goals of these projects I developed two new computational methods that can be utilized more broadly for the downstream interpretation of results from enrichment analyses of whole transcriptome profiles. There are a combined visualization and annotation approach called the Constellation Map and the Leading Edge Metagene Detector that systematically consolidates functionally related genes from multiple sets representing highly enriched biological pathways and processes in the comparison of expression data of two biological phenotypes. The application of those computational approaches and tools in this dissertation enabled a better understanding of the biological mechanisms related to human vaccine response. The software packages developed are freely available for use by biological investigators across many fields.