Fig. 4
Change in z-score and specificity due to simulated maternal duplications for different analysis approaches. Each row displays the performance of the indicated analysis strategy. The left column plots ∆zdup as a function of maternal duplication size on chr21, with each plot having >10,000 simulated samples. The thick trace is a rolling median, and the top and bottom thinner lines are the 75th and 25th percentiles, respectively. In the right column, the impact on specificity of maternal duplications is shown, with the dashed lines indicating the position of 95% specificity for chr21. The calculations for specificity and the expected frequency of false positives are described in Methods. (*) The indicated false-positive rate is an aggregate measure based on the specificity and mCNV prevalence of chromosomes 13, 18, and 21. a Simple, b Robust, c Robust+Gaussian, d Z-correction, e Value filtering, f mCNV filtering