Evaluation of copy number variant detection from panel-based next-generation sequencing data

Targeted gene capture and next-generation sequencing (NGS) has been widely utilized as a robust and cost-effective approach for detecting small variants among a group of disease genes. Copy number variations (CNV) can also be inferred from the read-depth information but the accuracy of CNVs called from panel-based NGS data has not been well evaluated.

Copy number variations covering adequate exons on autosomes can be accurately detected using targeted panel sequencing data as using CMA. CNVs detected from sex chromosomes need further evaluation and validation. Except for exon-level deletion/duplication and CNV on sex chromosome, our data support the use of panel-based NGS data for routine clinical detection of pathogenic CNVs.

Sequencing data were acquired from patients underwent routine clinical targeted panel sequencing testing. Pathogenic CNVs detected from targeted panel sequencing data were evaluated using CNVs generated by two clinical accepted platforms, namely chromosome microarray analysis (CMA) and multiple ligation-dependent probe amplification (MLPA) as benchmarks. CNVkit was used in our study to call CNVs from sequencing data using read-depth information. CMA and MLPA tests were used to confirm and further assess the size and breakpoints of CNVs.

The size of CNVs detected using panel-based NGS data are over 300 kb. The sizes of CNVs detected are slightly larger (102.3% on average) using the NGS platform than using the CMA platform, and the size accuracy improved as the size of variants increases. The breakpoints of CNVs detected using NGS data are quite close (within 2.3% of margin) to the breakpoints detected by CMA. CNVs on sex chromosomes, however, are less concordant between NGS and CMA platforms.