Identification of copy number variants in whole-genome data using Reference Coverage Profiles

Area of Science:

• Genomics
• Bioinformatics

Background:

• Accurate DNA copy number identification from short-read sequencing data is technically challenging due to large file sizes and algorithmic limitations.
• Existing whole-genome sequencing (WGS) methods often fail to detect small copy number variants (CNVs), especially hemizygous deletions.

Purpose of the Study:

• To develop an efficient method for identifying CNVs in individual genomes.
• To improve the detection of deletions in WGS data by addressing sequence-specific coverage fluctuations.

Main Methods:

• Developed a highly compressed representation of depth of coverage (150-1000× compression).
• Constructed multi-genome Reference Coverage Profiles (RCPs) to model diploid coverage across the genome, accounting for technical variations.
• Applied hidden Markov model (HMM) segmentation to normalized coverage data for CNV detection.

Main Results:

• Achieved significant compression of sequencing data, facilitating large-scale analysis.
• Demonstrated improved sensitivity and specificity in detecting CNVs, including hemizygous deletions, in individual genomes.
• Validated the method on over 6000 high-quality genomes.

Conclusions:

• Reference Coverage Profiles (RCPs) coupled with HMM segmentation provide an efficient and accurate method for CNV detection in individual genomes.
• This approach enhances the analysis of personal genomes and is crucial for clinical-grade genome interpretation.
• Provided publicly available RCPs and analysis tools.