Genome-Wide Survey of Host Responses: Use of Computational Analysis to Classify Exposures and Extract Signatures of Unconventional Versus Common Viral Exposures; Conference paper

摘要

Exposures to many unconventional pathogenic agents result in flu- like illness that are initially indistinguishable from common respiratory illnesses and early diagnosis to distinguish among the severe vs common viral infections depends on pathogen proliferation to dangerous, near-untreatable levels. Assessing exposure to a pathogen, in advance of onset of illness or at various stages post-exposure, is invaluable among the diagnostic options. Lymphocytes serve as 'whistle blower' indicators as they encounter pathogenic agents even early during the course of infection, registering the encounters in their mRNA and developing patterns of expression that correspond to each specific pathogen. Time series of gene expression patterns relate to the stage or severity of the infection and are unique for each pathogen. We are using the host blood for determination of whole genome regulation in response to various viral agents to extract features and signatures that can be used for point-of- care diagnosis of various viral infections (common respiratory, arena, flavi-, alpha- and other viruses). These data also have the potential to provide stage- appropriate therapeutic targets. These studies utilized exposure time sequences of host gene expression. The 'training sets' were constructed from in vitro exposures to purified peripheral blood leukocytes from approximately 6 human leukapheresis donors for each virus described above. Numerous modeling / mathematical techniques were applied to these datasets in order to identify signature patterns indicative of each. The 'shrink/grow' modeling approaches were used as well as other algorithms that have shown success for signature extraction. For the 'grow' algorithm, genes are individually selected that have the best discriminating power and the first of those frequently show properties unique for specific viral infections.