Background: Targeted next-generation sequencing (NGS) is increasingly being adopted in clinical laboratories for genomic diagnostic tests. Results: We developed a new computational method, DeviCNV, intended for the detection of exon-level copy number variants (CNVs) in targeted NGS data. DeviCNV builds linear regression models with bootstrapping for every probe to capture the relationship between read depth of an individual probe and the median of read depth values of all probes in the sample. From the regression models, it estimates the read depth ratio of the observed and predicted read depth with confidence interval for each probe which is applied to a circular binary segmentation (CBS) algorithm to obtain CNV candidates. Then, it assigns confidence scores to those candidates based on the reliability and strength of the CNV signals inferred from the read depth ratios of the probes within them. Finally, it also provides gene-centric plots with confidence levels of CNV candidates for visual inspection. We applied DeviCNV to targeted NGS data generated for newborn screening and demonstrated its ability to detect novel pathogenic CNVs from clinical samples. Conclusions: We propose a new pragmatic method for detecting CNVs in targeted NGS data with an intuitive visualization and a systematic method to assign confidence scores for candidate CNVs. Since DeviCNV was developed for use in clinical diagnosis, sensitivity is increased by the detection of exon-level CNVs.
Bibliographical noteFunding Information:
This study was supported by the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea [A120030] and the National Research Foundation of Korea grant funded by the Ministry of Education, Science and Technology, Republic of Korea [NRF-2017R1E1A1A03070512]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Funding for open access charge: The National Research Foundation of Korea.
© 2018 The Author(s).
All Science Journal Classification (ASJC) codes
- Structural Biology
- Molecular Biology
- Computer Science Applications
- Applied Mathematics