Effect of various transformations of -value prior to application of CBS on the power to detect a CNV.

Continuous outcome, 10,000 replications per cell, CNV size = 30 markers. “Power” here refers to the probability that a segment in which low -values have aggregated can be separated from the test results from surrounding markers. The “no pooling” analysis (which implements a Bonferroni correction to maintain the correct overall type I error rate) is included to illustrate the power gained by pooling information across nearby markers.