SRBreak: A Read-Depth and Split-Read Framework to Identify Breakpoints of Different Events Inside Simple Copy-Number...

Area of Science:

• Genomics
• Bioinformatics
• Population Genetics

Background:

• Copy-number variation (CNV) is linked to complex diseases.
• Detecting CNV breakpoints is crucial for understanding genome structure and evolution.
• Existing single-method tools struggle with accurate breakpoint identification.

Purpose of the Study:

• To develop a robust pipeline for identifying CNV breakpoints using high-throughput sequencing (HTS) data.
• To integrate read-depth and split-read information for improved breakpoint detection accuracy.
• To enable reliable breakpoint identification even for low-coverage and single-end sequencing data.

Main Methods:

• Developed a pipeline combining read-depth and split-read information from HTS data.
• Utilized a normal mixture model for sample clustering and an imputation approach.
• Employed kernel-based methods to maximize information from read-depth and split-read data, directly using CIGAR strings without re-alignment.

Main Results:

• The pipeline accurately identified approximate breakpoints (±10 bp) associated with ancestral events within CNVRs.
• Demonstrated effectiveness on simulated datasets, including very low-coverage and single-end read data.
• Achieved high concordance (92%, 100%, 82%) with the 1000 Genomes Project for human resequencing data at NEGR1, LCE3, and IRGM loci.

Conclusions:

• The developed pipeline offers a reliable method for CNV breakpoint detection by integrating multiple data types.
• The approach enhances the understanding of genome structure and evolution.
• The SRBreak package and Docker application are publicly available for broader research use.