Atlas-CNV: a validated approach to call Single-Exon CNVs in the eMERGESeq gene panel

• DOI: 10.1101/422337
• Publication Date: 2018-09-27T09:55:21Z
• Authors: Theodore Chiang; Xiuping Liu; Tsung-Jung Wu; Jianhong Hu; Fritz J. Sedlazeck; Simon White; Daniel Schaid; Mariza de Andrade; Gail P. Jarvik; David Crosslin; Ian Stanaway; David S. Carrell; John J. Connolly; Hakon Hakonarson; Emily E. Groopman; Ali G. Gharavi; Alexander Fedotov; Weimin Bi; Magalie S. Leduc; David R. Murdock; Yunyun Jiang; Linyan Meng; Christine M. Eng; Shu Wen; Yaping Yang; Donna M. Muzny; Eric Boerwinkle; William Salerno; Eric Venner; Richard A. Gibbs
• Keywords: DNA Copy Number Variations; Genome, Human; High-Throughput Nucleotide Sequencing; Humans; Exons; Sequence Analysis, DNA; Article; Software

Abstract Purpose: To provide a validated method to confidently identify exon-containing copy number variants (CNVs), with low false discovery rate (FDR), in targeted sequencing data from clinical laboratory particular focus on single-exon CNVs. Methods: DNA sequence coverage are normalized within each sample and subsequently exonic CNVs identified batch of samples (midpool), when the target log 2 ratio median exceeds defined thresholds. The quality CNV calls is assessed by C-scores (Z-like scores) using thresholds derived gold standard simulation studies. We integrate an ExonQC threshold lower FDR compare performance alternate software (VisCap). Results: Thirteen were used as truth set validate Atlas-CNV compared VisCap. demonstrated reduction validation, 10,926 eMERGESeq without sensitivity loss. Sixty-four multi-exon 29 high MLPA. Conclusions: generated laboratory. C-score assignment can reduce (identification targets variance) improve calling accuracy respectively. proposed guidelines criteria confidence