Figure 1From: Learning smoothing models of copy number profiles using breakpoint annotationsBreakpoint annotations quantify the accuracy of a smoothing model. Model agreement to annotated regions can be measured by examining the positions of predicted breakpoints b ̂ λ (vertical black lines) observed in the smoothing model ŷ λ (green lines). Black circles show logratio measurements y plotted against position p for a single profile i=375. Chromosomes are shown in panels from left to right, and different values of the smoothing parameter λ in the flsa model are shown in panels from top to bottom. Models with too many breakpoints ( λ=0.5) and too few breakpoints ( λ=10) are suboptimal, so we select an intermediate model ( λ=7.5) that maximizes agreement with the annotations, thus detecting a new breakpoint on chromosome 7 which was not annotated.Back to article page