biomvRhsmm: Genomic Segmentation with Hidden Semi-Markov Model
Figure 1
ROC curves for performance comparison. Receiver operating characteristic (ROC) curves for segmentation algorithm comparison under different signal-to-noise settings . Curves were generated by measuring the sensitivity and specificity at different threshold levels. The -axis and -axis show the false-positive rate (FPR) and true-positive rate (TPR), respectively. The upper panel (a) shows simulation 1, similar to an aCGH analysis, and the lower panel shows simulation 2, similar to peak identification using NGS. Compared algorithms are color-coded as indicated in the figure legend, while the up-triangle represents segment of gain in simulation 1 and peak in simulation 2, and hollow down-triangle represents segment of loss in simulation 1. Models are labeled using lowercase letters of their name. Our proposed model is coded as “hsmm” for simplicity and the hidden Markov model in package aCGH is labeled as “hmm.”
