Abstract:
|
The promise of gene microarrays is that the technology could allow researchers to accurately quantify expression in RNA levels of thousands of genes in a tissue sample, thereby providing valuable information about complex gene expression patterns. Recent advances in bioinformatics have brought us closer to realizing this promise. However, the massive scale and variability of microarray gene data creates new and challenging problems of signal extraction, gene clustering, and data mining, especially for temporal studies. Almost all methods introduced for discovering differentially expressed genes are based on thresholding a single discriminant, e.g. a ratio of between-class to within-class variation. In this talk we introduce a different approach for extracting information from gene microarrays which is based on a novel statistical formulation of multi-objective optimization. We will illustrate our methods by applying it to a GeneChip study for detection of genes expressed in the mouse retina which are associated with aging.
This work is in collaboration with Gilles Fleury (ESE Paris).
|