Figure 4

ROC comparison with scaled data (Prostate Cancer dataset, sample GSM318766, hybridized to Mapping250K_Nsp). We have focused on the region where there is a change from 1 to 2 copies, around position 32Mb, that appears in Figure3. For the analysis, we have included the SNPs that surrounds the copy number change (specifically from 20 to 45 Mb). The SNPs within a safety zone from 30Mb to 34 Mb are not considered in the analysis, because it is difficult to discern the exact position of the change. The SNPs located upstream the change point are considered to have total CNs equal to 1 (deleted region) and the SNPs downstream the change point to have a normal CN (CN=2). For this particular sample (in the studied jump) the results using NSA are almost as good as using MCS. MTS gives poorer results and MHS is well behind. This is a general trend that discourage the joint analysis of samples from different laboratories.