Microarray Data Analysis: General Concepts, Gene Selection, and Classification

Disclosures from the genome sequencing projects facilitated the development of novel techniques able to screen thousands of molecules in parallel and identify sets of potentially interesting sequences associated with physiological/pathological conditions. As a consequence, high-throughput, large-scale experimental methodologies, combined with bioinformatics analysis of DNA, RNA, and protein data, projected biological sciences into the so-called post-genomic-functional genomics era. The exploration of all genes or proteins at once, in a systematic fashion, represents a sort of revolution, shifting molecular biology and medicine research from a reductionistic, hypothesis-driven approach towards deciphering how genes and their products work, how they interact in pathways within the cells, and what roles they play in health and disease [1-2]. Oligonucleotide and cDNA microarrays for transcriptional profiling (Lockhart et al. 1996, Schena et al. 1995) allow measuring such interaction patterns, thus representing an unprecedented opportunity to boost the identification of diagnostic and therapeutic targets (Brown et al. 1999).The principle of a microarray for gene expression analysis is basically that of the classical northern-blot extended to the whole genome level. Specifically, mRNA from a given cell line or tissue is labeled with a fluorescent dye and hybridized to a large number of DNA sequences, immobilized on a solid surface (i.e., a glass slide or a silica wafer) in an ordered array. In such a way, tens of thousands of transcript species can be detected simultaneously, exploiting the highly specific complementary of nucleic acids. Indeed, mRNA molecules (targets) will couple with the corresponding complementary probe immobilized on the array and the fluorescence emission of any spot in the array will represent a quantification of the expression level of each target