Abstract
Purpose :
With recent advances in computational tools, copy number variation (CNV) calling based on next-generation sequencing (NGS) data has emerged as a promising alternative to the current gold standard of array comparative genomic hybridization (aCGH). Through a high-coverage, targeted-panel approach, we set out to validate the clinical utility of NGS-based CNV calling as well as outline its dependencies and limitations.
Methods :
Targeted sequencing of 536 ophthalmology related genes was performed on a cohort of 512 retinal dystrophy patients with an average sequencing depth of 622+/-117X across the panel. Sequencing was done using Illumina’s HiSeq2500 and the data processed and analyzed using GATK best practices and the eXome-Hidden Markov Model (XHMM) CNV caller. Control samples with known copy number status at specific loci were identified through gel analysis and TaqMan qPCR. TaqMan qPCR was also used to confirm CNVs identified by the NGS CNV pipeline.
Results :
A total of 325 CNVs were found across 7406 coding exon targets for the 484/512 samples passing QC with an estimated false positive rate of 6.1%. Of the 51.3% of clinical cases negative or inconclusive from sequencing without CNV analysis, a clinically relevant, disease causing CNV was identified and confirmed in 12.1%, improving overall diagnostic yield of the panel from 48.7% to 54.9%. Sensitivity and specificity of CNV callings were both positively correlated with and highly dependent upon sequencing depth and size of the CNV. The median sequencing depth of a called CNV was 1176X spanning a median of 1.58kb and 2 exon targets. The median sequencing depth of a clinically relevant, causative CNV was 766X spanning a median of 2.54kb and 2.5 exon targets. The minimum sequencing depth required to call a true-positive, single-exon CNV was 781X and the smallest, confirmed, single-exon CNV target called was 97 base pairs.
Conclusions :
With the ability to call CNVs at a resolution finer than aCGH, NGS-based CNV calling presents itself as an effective screening method for clinical diagnostics without the need for additional molecular work. This extension of NGS data analysis should be implemented in any high coverage sequencing application to supplement primary sequencing findings and significantly improve mutation detection rates.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.