Table 2. Major features of programs for detecting copy number variation in cancer genome using next generation sequencing data¹.

Programs	Data type	Data preprocessing3	Segmentation	Interpretation	Sample information
SegSeq	RC	Matched normal;	Local change-point analysis with a subsequent merging procedure	Optimized cutoffs	/
ReadDepth	RD Discordant read pairs	Mappability correction; GC correction; RD Negative-binomial distribution	CBS	Optimized cutoffs	/
BIC-seq	RD	Matched normal; No data distribution assumption	Minimizing BIC	Empirical cutoffs	/
Patchwork	RD BAF	Normal genome; GC correction	CBS	Pattern Recognition and empirical cutoffs	Tumor purity Tumor ploidy
OncoSNP-SEQ	RC BAF	Matched normal; Mappability correction; GC correction Mixture of uniform and binomial distribution	HMM	HMM	Tumor purity Tumor ploidy Tumor heterogeneity
HMMcopy	RC	Matched normal; Mappability correction; GC correction	HMM	HMM	/
CONSERTING	RD BAF Soft-clipped reads	Matched normal; Mappability filtering; GC correction	Regression Tree	Empirical cutoff	/
ExomeCNV	RD2	Matched normal	CBS	Optimized cutoff	Fixed tumor purity
VarScan2	RD	Matched normal	CBS	Empirical cutoff	/
HAPSEG/ABSOLUTE	RD at SNP loci	Matched normal	probabilistic method	Pattern Matching and fit platform error model	Tumor purity Tumor ploidy Existence of sub-clone
Control_FREEC	RC	Matched normal and/or GC and Mappability correction;	LASSO algorithm	Empirical cutoff	Tumor purity User inputs tumor ploidy

¹Abbreviations: RC, Read Counts; RD, Read Depth; BAF, B Allele Frequency; SNP, single nucleotide polymorphism; CBS, circular binary segmentation; HMM, hidden Markov model.

²ExomeCNV uses only RD for calling CNV; it uses BAF for calling LOH.

³The data is assumed to be in normal distribution if not specified.