ABSTRACT
The widespread use of DNA microarray gene expression technology offers the potential to substantially increase our understanding of cellular pathways and apply that knowledge in clinical as well as basic biological studies. In human biology, there are two common types of study performed with expression arrays. The first involves samples derived from cloned cells in a highly controlled environment, subject to some designed experimental intervention, such as up/down regulation of a particular target gene (Huang, West and Nevis , 2003; Huang, et al., 2003; Bild, et al. , 2006) or manipulation of the pH level of the growing environment (Chen, et al. , 2008). Such experiments are useful in that they are designed to minimize all sources of gene expression variability except that of the experimental intervention. One result generated by this type of experiment is the identification of a subset of genes that are differentially expressed when subject to a given intervention. A list of such genes, together with the magnitudes by which those genes are differentially over/under-expressed and additional numerical summaries are said to comprise the signature of the intervention. The second, common type of study involves samples derived from tissue removed from living subjects, such as tumor tissue obtained via biopsy (Perou, et al., 2000; West, et al. , 2001; Huang, et al., 2002, 2003; Pittman, et al. , 2004; Seo, et al., 2004; Miller, et al. , 2005; Riech, et al., 2005). A central goal of this second type of study is to
find statistical associations between the measured expression levels in a tissue sample and clinical variables associated with the subject, e.g. survival time or metastasis.
