Filtered CNV counts stratified by copy-number state before and after outlier removal.

The table summarizes the filtered CNV counts by copy-number state before and after outlier removal. A group of 77 cases (3.6%) and 351 controls (2.0%) percentage of samples were found to have very high CNV-counts (>6). Visual inspection of many of these CNVs revealed that a majority of these CNVs are false positives and that these samples generally have 1–3 true CNVs. Based on this observation we applied an outlier-removal filter to minimize the inflation of CNV-counts caused by sample specific and systematic effects. The frequency of each CN state is shown in parenthesis. After outlier removal the frequency of the different CN-states are quite similar in the case and control populations.