Discriminant Analysis using Multigene Expression Profiles in Classification of Breast Cancer

摘要

Microarrays provide a promising tool for the diagnosis of cancers on a molecular level. However, due to the large dimensions and the intrinsic variations of such data, feature selection that reduces the data to a small number of "informative genes'' is important before any discriminant analysis based on such data. Many marginal statistical measures have been applied to gene expression data despite the fact that gene-gene interactions are not negligible. In this paper, we propose a gene selection procedure, the multigene profile association score (MPAS) method, based on backward screenings using multigene association information that captures interactions. As a result, not only genes with marginal significance are detected, but also those contain information on interactions that are only detectable when being evaluated together with other genes.