Evaluation of CNV detection tools for NGS panel data in genetic diagnostics

Area of Science:

• Genetics
• Bioinformatics
• Medical Diagnostics

Background:

• Germline copy-number variants (CNVs) are established causes of hereditary diseases.
• Accurate detection of CNVs, especially single and multi-exon alterations, from targeted next-generation sequencing (NGS) data remains a significant challenge in genetic diagnostics.

Purpose of the Study:

• To evaluate the performance of different CNV calling tools using targeted gene panel NGS data.
• To assess the suitability of these tools as a screening step prior to orthogonal confirmation in genetic diagnostic workflows.

Main Methods:

• Five CNV calling tools (DECoN, CoNVaDING, panelcn.MOPS, ExomeDepth, CODEX2) were tested.
• Evaluation was performed on four genetic diagnostic datasets (2 in-house, 2 external) comprising 495 samples with 231 validated single and multi-exon CNVs.
• Tools were assessed using both default and optimized parameters for sensitivity.

Main Results:

• Most evaluated tools demonstrated high sensitivity and specificity, though performance varied across datasets.
• DECoN and panelcn.MOPS exhibited strong performance in the diagnostic scenario, detecting nearly all validated CNVs.
• DECoN achieved superior specificity (>0.90) with optimized parameters compared to panelcn.MOPS.

Conclusions:

• DECoN and panelcn.MOPS are highly effective for screening single and multi-exon CNVs in targeted NGS data for genetic diagnostics.
• Optimized DECoN offers excellent sensitivity and specificity, making it a valuable tool for pre-confirmation CNV screening.
• The study provides a benchmark for CNV calling tools in diagnostic settings, aiding in the selection of appropriate methods.