Figure 3

Schematic representation of the gene-set enrichment procedure followed. (a) The CNAs are mapped onto the chromosomal coordinates and then used to (b) build up a ranked list of cytobands, which is further used to (c) define a ranking of genes according to how frequently are they involved in CNAs. The distribution of gene modules, in this case the annotations of a fictitious GO, derived from such ranked list is tested for its significant accumulation in frequently lost regions. (d) The variant of the GSA test used [14] seeks for significant asymmetrical distributions of annotations (red bars) with respect to the annotation background (blue bars). The × axis in d represents the number of cases in which a gene has been observed in region affected by a CAN (that, generally speaking can be either amplification or a loss). The bar height represents the number of genes involved in a given frequency of CNA events. The red bars represent the distribution of the genes of a given GO across the frequency of observations of CNAs. The GSA test seeks for GOs whose distributions are significantly skewed towards high or low frequencies of CNAs. Since a large number of GO terms are tested, the FDR method [34] was used to adjust the p-values for multiple-testing effects. The Mitelman database of Chromosome Aberrations in Cancer http://cgap.nci.nih.gov/Chromosomes/Mitelman, May 2007 release, was used as primary source of information. A total of 86048 observations of deletions (corresponding 19859 to regional deletions and 66189 to whole chromosome deletions), and 55935 observations of amplifications (corresponding to 1011 regional amplifications and 54924 whole chromosome amplifications) including any type of cancer, were obtained from the database.