SoftSearch: integration of multiple sequence features to identify breakpoints of structural variations

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Structural variations (SVs) are a significant component of genetic makeup, yet remain poorly understood.
• Next-generation sequencing (NGS) technologies are crucial for SV discovery, but a standardized detection tool is lacking.

Purpose of the Study:

• To develop and validate SoftSearch, an algorithm for discovering structural variant breakpoints in Illumina paired-end NGS data.
• To address the need for a reliable and comprehensive SV detection tool.

Main Methods:

• SoftSearch integrates multiple SV detection strategies: split-read, discordant read-pair, and unmated pairs.
• Co-localized split-reads and discordant read pairs are utilized to refine breakpoint identification.
• The algorithm does not require secondary alignment and is portable into existing sequencing workflows.

Main Results:

• SoftSearch was validated using both real and synthetic datasets.
• The algorithm identifies breakpoints using minimal evidence from soft-clipped bases and discordant read-pairs.
• SoftSearch is applicable to diverse DNA sequencing experiments, including whole genome and exome sequencing.

Conclusions:

• SoftSearch enhances true SV identification by combining multiple sequence features.
• The tool successfully identified clinically relevant SVs in the BRCA2 gene, which were missed by other methods.
• SoftSearch demonstrates significantly improved overall performance in SV detection.