Understanding the dynamics of host-virus interactions is critical for developing effective antiviral strategies. Our long term goal is to develop a quantitative method-based on signal processing techniques-to derive a characteristic profile of multiple host-virus interactions in vitro. The significance of such a profile lies in its ability to indicate important features such as: the rate of viral replication; the effectiveness of cell-based innate immune responses; and, the ability of infected cells to induce an antiviral state in the local, uninfected cells. The model system will be influenza A (A/WSN/33 and A/Memphis/76) infecting human HEK293 and A549 cells. The specific problem driving our proposed research involves the identification, collection, denoising, and analysis of informative marker signals that will be used to develop the profile. Our approach to deriving the profiles has three innovative aspects: (1) the use of immunofluorescent images to generate the signals that quantify the complex phenotype; (2) the examination of host and virus signals in both space and time (with the spatial resolution providing new information that is not available in many standard biochemical assays); and, (3) the design of signal and image processing techniques to extract the principal information from the experimental assay. The following three specific aims will demonstrate the feasibility of our approach for developing the characteristic profile of a virus; they will also provide initial evidence of the efficacy of the profiles in quantitatively characterizing complex phenotypes. 1. Identify informative markers based on functional genomics data and generate images. Preliminary microarray experiments of host responses in the two cell lines and mice infected with flu will provide a starting point for the selection of appropriate markers associated with all stages of the infection process. Cell monolayers will then be infected and immunostained for the spatial and temporal behavior of those markers. 2. Generate high-quality immunofluorescent intensity signals (IIS) from noisy images. The montaged images must first be corrected to eliminate artifacts that arise due to low magnification imaging and to normalize intensity disparities that are introduced during acquisition. We will design an automated, optimal denoising method that generates higher quality images, significantly faster, than the current manual system. 3. Analyze signals and design a method to derive the host-virus profile. The assay data represent a very large collection of high-quality signals; each signal denotes a marker in space (radial propagation distance from the initial infection site) and time (elapsed time from initial infection). We will employ signal processing methods to identify and extract a set of """"""""sufficient features"""""""" from this myriad data. The result will be a profile that accurately models imoortant host-virus interactions.
