Fig. 6

Comparison of CNV calling detection time in a large cohort of samples using “kNN” and “k-means” based approach to reference selection. The time measured includes reference set selection, normalization and CNV calling step. Note that CNV calling is performed for all samples in the cohort. The substantial difference between total execution times is a consequence of normalization process as required by “kNN” method (once per sample) and “k-means” approach (once per cluster). Since the number of samples is significantly greater than the number of groups, the total normalization time is significantly greater for “kNN” than for “k-means”