High-quality, quantitative microarrays are an absolute requirement of array experiments involving comparative analysis of related genomes, or sensitive and specific diagnosis of infectious disease. Our goal is to develop a suite of software applications that incorporate experimentally validated solution hybridization parameters into the processes of microarray design and microarray data preprocessing. The specific hypothesis we propose to test is that, using solution hybridization parameters, we can accurately predict the binding behavior of perfectly matched and several classes of mismatched oligonucleotides on an array.
Specific aims of the project proposed herein are 1) to model and predict hybridization behavior of DNA arrays based on solution hybridization parameters, 2) to design test arrays to quantify the impact of biophysical properties of the probe and target on array hybridization and assess our ability to predict hybridization, 3) apply biophysical criteria to a real-world probe design problem and use the experience to develop a best-practice approach for using these criteria in design, 4) integrate biophysical modeling with accepted array design and analysis procedures in an automated pipeline, and 5) develop a user interface for the software and tools for integration of biophysics-based analysis with established microarray data analysis procedures. The predictive models that we develop can be applied both in the design of optimized arrays and in preprocessing of raw signal for analysis. As an empirical test of our methods and array designs we will supply versions of a diagnostic array to discriminate among sequenced species of Brucella. Brucella is classified by the CDC as a class B pathogen;as an intracellular pathogen it has the alarming ability to infect macrophages and to be masked from the host's immune response. We will provide a well-designed and tested oligonucleotide microarray with the necessary probes for discrimination among three biovars of Brucelia (abortus, melitensis, and suis).
