Fig. 1

Results of the transformation of all copy number segments from the entire dataset (13 clinical and simulated CNVs). The transformation procedure is used to change the scale of \(\log _2\) copy number values for subsequent removal of all non-CNV segments using a K-means clustering algorithm. A Distribution of segments’ mean values before the transformation. In this distribution, there is no clear separation between normal segments (i.e., those that correspond to the diploid state), which form a peak around zero, and potential CNVs (tails of the distribution). B Segments’ mean and standard deviation values after the transformation. Circles and triangles denote segments corresponding to real CNVs from the dataset. After applying clustering on the transformed values, it becomes easy to remove false positives (the blue and green clusters) while keeping all true-positive CNVs (the red cluster, centered around 0)