In addition we required that all CNVs be reproduced by 3 independent programs, a step that increases confidence in the results but that raises the potential problem of the same CNV region being detected with different starting and ending points, which results in uncertainty on how to combine these different CNV calls. In order to avoid this controversy, we adopted an intuitive method where we tested each genetic marker instead of a particular CNV segment. A CNV status is assigned to a particular genetic marker when all programs report a CNV that covers this probe (Supplementary Tables 1 & 2). We validated our findings by qPCR – while the region on chromosome 5q13.2 was successfully validated, the duplication on 6q14.1 did not. Thus, the finding on 6q14.1 should we viewed with caution and as such, it cannot be claimed as a true CNV until laboratory validation is successful. However, we cannot be certain that absence of experimental validation necessarily negates a CNV called using statistical algorithms – in some instances, lack of suitable laboratory probes in the region of the
