Fig. 5From: Comparison of kNN and k-means optimization methods of reference set selection for improved CNV callers performanceComparison of CNVs detected by CODEX with four different selection methods. The k value for “k-means” method was set to 4, and k for “kNN” algorithm was fixed at 200; the baseline (“all samples” method) corresponds to the value of “k-means” for k = 1. The figure shows, that using “kNN” and “k-means” methods results in better precision in comparison to the baseline. What is more, sensitivity for all of the investigated methods remains fairly stable. It is worth noting that the dots for the “kNN” and “k-means” methods are very close to each other - both of mentioned methods lead to very similar resultsBack to article page