Description of the SIRAC algorithm. Illustration of the algorithmic steps of SIRAC. The corresponding results for the NKI dataset are shown. The data is labeled according to the cancer subtypes introduced by Sorlie and Perou [30, 31, 32]; in this example the label Luminal A subtype versus all others subtypes is used. In Step 1 the relevant DNA-probes are selected. Each DNA-probe is plotted on the genomic location with two circles of different color representing the median value of the samples in the two classes. In Step 2, the vertical axis represents the different window sizes, the blue lines along the genome (the horizontal axis) show the regions judged significant by the algorithm. In the final step, Step 3, the number of window sizes for which the location is judged significant by the hyper-geometric test are shown along the vertical axis. The relevant region selected when s = 9 is highlight by the red curve.