Detecting copy number variation with mated short reads.

Area of Science:

• Genomics
• Bioinformatics
• Human Genetics

Background:

• High-throughput sequencing (HTS) enables novel copy number variant (CNV) detection methods.
• Previous HTS-based CNV detection methods are limited by sequencing biases affecting resolution and breakpoint accuracy.
• Array comparative genomic hybridization (aCGH) was a traditional method for CNV detection.

Purpose of the Study:

• To develop a new computational method for accurate CNV detection using HTS data.
• To mitigate sequencing biases inherent in HTS data for improved CNV identification.
• To validate the performance of the new method against existing databases and known variants.

Main Methods:

• Developed CNVer, an algorithm integrating depth-of-coverage and paired-end mapping information from HTS data.
• Utilized a "donor graph" computational framework to unify and analyze these data types.
• Applied CNVer to detect CNVs in a Yoruban individual's genome.

Main Results:

• CNVer identified 4879 CNVs in the Yoruban genome.
• 77% of CNVer calls overlapped with known variants in the Database of Genomic Variants.
• 81% of known deletion CNVs for the individual were identified by CNVer.
• Demonstrated CNVer's ability to reconstruct absolute copy counts and its feasibility with low-coverage data.

Conclusions:

• CNVer accurately detects CNVs by combining HTS depth and mapping information, effectively mitigating sequencing biases.
• The algorithm shows high concordance with existing variant databases and known CNVs.
• CNVer is a promising tool for CNV detection, even with low-coverage sequencing data.