A Web-based data warehouse on gene expression in human malignant melanoma.

The identification of melanoma-specific dysregulated genes could identify new molecular markers. By applying bioinformatic tools for screening of biomedical databases, a melanoma-specific gene expression profile "data warehouse" was constructed. Utilizable data sets of global gene expression analyses were available from nine studies that applied different technology platforms. A single study used cell lines, five investigations analyzed cell lines and tissues obtained from patients, two studies used exclusively specimens obtained from patients, and one study analyzed blood cells prepared from patients. The total number of investigated patients was 116. From 815 differential-regulated genes, 772 (95%) were identified merely in a single study, 37 in at least two studies, five (RAB33A, ERBB3, ADRB2, MERTK, SNF1LK, and ITPKB) in at least three studies, and a single gene, RAB33A, in four studies. These data show that the accuracy, reproducibility, and comparability among different gene expression profile studies are low in melanoma. In conclusion, the study demonstrates the high diversity of gene expression profiles associated with melanoma, the necessity to include a sufficient number of samples regarding clinical standards, for the design of standardized sample collecting and preparation, for the development of common standards for microarray data processing, and for developing standardized bioinformatic tools.