Fig. 5

Effect of the size of the set of sequencing regions which models background on the results’ sensitivity and precision. a, b relate to the selection of the k most correlated sequencing regions, c, d k random sequencing regions. Additionally, b, d the improvement of individual results relative to the baseline, i.e. all sequencing regions are normalized simultaneously. It is worth paying attention to the fact that reducing the value of k in the knn algorithm to a value equal to 100 (b) allowed for a 3-fold improvement in precision while maintaining a constant level of sensitivity for rare CNVs