Comprehensive evaluation of structural variation detection algorithms for whole genome sequencing

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Structural variations (SVs) and copy number variations (CNVs) significantly influence gene function and are implicated in various human diseases.
• Current whole genome sequencing (WGS) data analysis tools struggle to detect all SV types with high precision and recall.

Purpose of the Study:

• To comprehensively evaluate the performance of 69 existing SV detection algorithms.
• To identify optimal algorithms for different SV types and sizes.
• To assess the impact of combining algorithms on SV detection accuracy.

Main Methods:

• Performance evaluation of 69 SV detection algorithms using simulated and real WGS datasets.
• Systematic assessment of overlapping calls between algorithm combinations for various SV types and sizes.
• Analysis of precision, recall, breakpoint, size, and genotype accuracy.

Main Results:

• A subset of algorithms demonstrated high accuracy for specific SV types and size ranges.
• GRIDSS, Lumpy, SVseq2, SoftSV, Manta, and Wham were identified as effective for deletion and duplication detection.
• Combining specific algorithm pairs, rather than general methods, improved precision and recall for SV calling.

Conclusions:

• Careful selection of SV detection algorithms tailored to specific SV types and sizes is essential for accurate genomic analysis.
• Leveraging overlapping calls from carefully chosen algorithm pairs offers a promising strategy to enhance SV detection accuracy.