Dysgu: efficient structural variant calling using short or long reads

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Structural variations (SVs) are crucial in genome evolution and disease, including cancer.
• Long-read sequencing enhances SV characterization, but paired-end sequencing offers scalability.
• Accurate detection of SVs and indels remains a challenge in genomic analysis.

Purpose of the Study:

• To introduce dysgu, a novel tool for detecting structural variations (SVs) and indels.
• To evaluate dysgu's performance using both paired-end and long-read sequencing data.
• To compare dysgu against existing state-of-the-art SV detection tools.

Main Methods:

• Dysgu analyzes alignment gaps, discordant, and supplementary mappings.
• It generates consensus contigs and employs machine learning for event classification.
• Remapping of anomalous sequences aids in identifying additional SVs.

Main Results:

• Dysgu demonstrates superior performance over existing tools for both paired-end and long-read data.
• The tool achieves high sensitivity and precision in SV and indel detection.
• Dysgu is among the fastest available tools for SV calling.

Conclusions:

• Dysgu provides a scalable and accurate solution for structural variation and indel detection.
• Combining low-coverage paired-end and long-reads offers a competitive alternative to high-coverage long-reads.
• Dysgu advances the field of genomic variation analysis, with implications for disease research.