Copy number variants (CNV) are a major cause of disease, with over 30,000 reported in the DECIPHER database. To use read depth data from targeted Next Generation Sequencing (NGS) panels to identify CNVs with the highest degree of sensitivity, it is necessary to account for biases inherent in the data. GC content and ambiguous mapping due to repetitive sequence elements and pseudogenes are the principal components of technical variability. In addition, the algorithms used favour the detection of multi-exon CNVs, and rely on suitably matched normal dosage samples for comparison. We developed a calling strategy that subdivides target intervals, and uses pools of historical control samples to overcome these limitations in a clinical diagnostic laboratory. We compared our enhanced strategy with an unmodified pipeline using the R software package ExomeDepth, using a cohort of 109 heterozygous CNVs (91 deletions, 18 duplications in 26 genes), including 25 single exon CNVs. The unmodified pipeline detected 104/109 CNVs, giving a sensitivity of 89.62% to 98.49% at the 95% confidence interval. The detection of all 109 CNVs by our enhanced method demonstrates 95% confidence the sensitivity is ≥96.67%, allowing NGS read depth analysis to be used for CNV detection in a clinical diagnostic setting.
Bioinformatic Challenges in Clinical Diagnostic Application of Targeted Next Generation Sequencing: Experience from Pheochromocytoma
