The challenge of cancer treatment has been to target specific therapies to pathogenetically distinct tumor subtypes to maximize efficacy and minimize toxicity. However, tumors with similar histopathological appearance can follow significant different clinical courses and show different responses to therapy. The recent development of gene microarrays provides an opportunity to take a genome-wide approach to predict clinical heterogeneity in cancer treatment. Although such global views are likely to reveal previously unrecognized patterns of gene regulation and generate new hypotheses warranting further study, widespread use of microarray profiling methods is limited by the need for further technology developments, particularly comprehensive bioinformatics tools not previously included by the instruments. The long-term goal of the proposed work is to develop, test, and disseminate effective bioinformatics tools to interpret the rich information derived by gene microarrays about underlying cancer biology (e.g., molecular biomarkers) and to facilitate molecular classification/prediction of cancer and response to therapy. This technology-driven project is inspired by the underlying hypothesis that microarray based gene expression profiling and integrated intelligent bioinformatics tools can, at the molecular level, (1) confirm existing and discover previously unrecognized cancer phenotypes; (2) identify most relevant diagnostic or therapeutic biomarkers; and (3) predict diagnosis, prognosis, and response to therapy. The R21 project will focus on: (1) performing rigorous and quantitative tests to compare the proposed methods with comparable existing methods, and (2) performing quantitative tests to show the feasibility of the proposed methods where no comparable method exists. The R33 project will focus on: (1) establishing a database of gene expression profiles derived from two human cancers (breast & childhood tumors); (2) extracting and refining most relevant biomarkers associated with previously and newly defined cancer phenotypes; and (3) developing, optimizing, and validating neural network classifiers to predict tumor phenotype and response to therapy with confidence values. These novel bioinformatics tools will be developed based on state-of-the-art and/or latest inventions in engineering, computer science, advanced statistics, and neural networks, and will produce a major advance in the molecular analysis of cancer. ? ?
Showing the most recent 10 out of 18 publications
