biomvRhsmm: Genomic Segmentation with Hidden Semi-Markov Model

Table 2

Area under the ROC curves of simulation 1 data.

	𝑟=1		𝑟=2		𝑟=3		Weighted avg. rank
	AUCg	AUCl	AUCg	AUCl	AUCg	AUCl
hsmm	0.619487	0.729668	0.982176	0.988689	0.999141	0.998519	3.528400
bcp	0.675283	0.758042	0.912009	0.956031	0.921090	0.963401	6.416511
bioHMM	0.685575	0.875202	0.977079	0.990215	0.995062	0.997551	3.408419
CBS	0.633272	0.795941	0.974008	0.985508	0.996065	0.995409	4.491084
CGHseg	0.586505	0.696329	0.960183	0.991765	0.996045	0.998051	4.621081
GLAD	0.506588	0.548763	0.833011	0.962862	0.986098	0.996577	6.907573
HaarSeg	0.649687	0.763682	0.923416	0.993653	0.995908	0.998408	3.886978
HMM	0.717595	0.854783	0.749492	0.887209	0.526358	0.573611	6.589848

AUC_g and AUC_l are area under the receiver operating characteristic (ROC) curves for simulated gain and loss segments, respectively, for each 𝑟.
Weighted avg. rank is calculated as 𝑛+1−∑^𝑗=𝑐_𝑗=1AUC_𝑖×rank^𝑗(AUC_𝑖)/𝑐 for each model 𝑖, where 𝑐 is the number of AUC columns and n is the number of competing models.