SVIM: structural variant identification using mapped long reads.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Structural variants (SVs) are large genomic alterations (>50 bp) impacting phenotype and disease.
• SVs affect more genomic bases than smaller variants and are challenging to detect with short-read sequencing.
• Long-read sequencing technologies offer advantages for SV detection but existing tools are suboptimal.

Purpose of the Study:

• To present SVIM, a novel computational tool for sensitive and precise detection of structural variants from long-read sequencing data.
• To enable robust characterization of various structural variant classes, including duplications and insertions.

Main Methods:

• SVIM employs a three-component approach: signature collection, clustering, and combination from read alignments.
• The tool analyzes long-read sequencing data generated by Pacific Biosciences and Oxford Nanopore Technologies.
• SVIM differentiates five distinct structural variant classes, including complex duplication types.

Main Results:

• SVIM demonstrates sensitive detection and precise characterization of structural variants.
• The tool effectively extracts genomic origin and destination information for duplications.
• Comparative evaluations show SVIM outperforms existing tools on both simulated and real long-read datasets.

Conclusions:

• SVIM is a powerful new tool for analyzing structural variants in long-read sequencing data.
• The software improves the detection and characterization of genomic variations linked to human diversity and disease.
• SVIM is readily available and implemented in Python 3.