false negative rate here (3.5%, 3/85 CNV calls), the adjustment procedure preserves signals from genuine CNVs in these CNV regions and does not introduce false positive calls. To further validate this visually, we plotted examples of signal intensities before and after adjustment for samples with experimentally validated CNVs (Supplementary Figure 8), and show that the signal adjustment procedure reduces the overall magnitude of genomic waves but has little effects on the signals of the true CNV regions. Table 2.Comparison of computationally generated CNV calls with Q-PCR results on six CNV regions on 48 subjects in a large pedigreeCNV regionAberrant copy numberNumber of subjects with CNVs by Q-PCRNumber of subjects with CNV calls (Number of overlapping with Q-PCR results)With signal adjustmentWithout signal adjustment4p1231916 (16)16 (16)4q13.2166 (6)6 (6)6q14.10, 12929 (29)29 (29)6q2741010 (10)10 (10)12p13.3131515 (15)15 (15)15q14166 (6)6 (6)The CNV calling results are the same before and after signal adjustment on these six regions, and only three false negative CNV calls were made, indicating a false negative rate of 3.5% without or with wave adjustment. In all cases, the exact copy numbers (0, 1, 3 or 4) for CNVs were identified correctly.
