Fig. 1

CNV detection workflow. The tumor and normal samples were first aligned to the human genome and processed independently to remove PCR duplicates and perform local realignment. Common SNPs were detected in the tumor sample using GATK [30]. CNV Radar then uses the alignment files from tumor and normal samples to calculate relative read depths (log2 fold change; black) at all capture regions. VAF scores (orange), which indicate the level of deviation from a copy neutral state, were calculated for all heterozygous common SNPs in the tumor sample. The smoothed log2 fold change and VAF scores were then combined into CNV scores, which CNV Radar then uses to iteratively detect breakpoints, re-center the genome-wide read depth, and make the final CNV calls (green). CNV, copy number variation; GATK, Genome Analysis Tool Kit; VAF, variant allele frequency