Improving CNV Detection Performance Except for Software-Specific Problematic Regions.

Area of Science:

• Genomics
• Bioinformatics
• Medical Genetics

Background:

• Whole exome sequencing (WES) is a powerful tool for identifying disease-causing genetic variants.
• Copy number variation (CNV) detection using WES data often suffers from limited sensitivity and high false-positive rates.
• Accurate CNV detection is crucial for diagnosing genetic disorders.

Purpose of the Study:

• To evaluate the performance of four WES-based CNV callers (CNVkit, CoNIFER, ExomeDepth, cn.MOPS).
• To develop a strategy for improving CNV detection sensitivity and specificity using WES data.
• To identify and filter problematic genomic regions that negatively impact CNV caller performance.

Main Methods:

• Constructed a reference CNV set using chromosomal microarray analysis (CMA) data.
• Evaluated four WES-based CNV callers against the CMA benchmark.
• Identified software-specific problematic genomic regions and filtered out overlapping CNVs.

Main Results:

• The four CNV callers exhibited low concordance and distinct problematic region distributions.
• An average of 1.23% of sequencing target baits were identified as problematic due to low mappability and high variation.
• Targeted filtration significantly improved the performance of all tested CNV callers, with ExomeDepth showing notable gains in sensitivity and positive predictive value.

Conclusions:

• Software-specific problematic regions in WES data can be delineated.
• Targeted filtration is an effective strategy to reduce false positives in WES-based CNV detection.
• Improved CNV detection accuracy using WES data can enhance genetic diagnosis.