Benchmark for simple and complex genome inversions.

Area of Science:

• Genomics
• Structural Variation Analysis
• Bioinformatics Tool Development

Background:

• Inversions are significant structural variations implicated in genomic disorders, evolution, and instability.
• Detecting inversions is challenging, especially in repetitive regions and for complex rearrangements.
• A lack of high-quality benchmarks hinders progress in inversion detection and interpretation.

Purpose of the Study:

• To create a comprehensive, multi-genome benchmark for evaluating inversion detection methods.
• To assess the performance of leading structural variant callers and alignment strategies across different sequencing platforms.
• To identify strengths and limitations of current tools for robust biological interpretation.

Main Methods:

• Developed a benchmark using Strand-seq and phased long-read assemblies across five reference samples.
• Refined breakpoints using haplotype-resolved long-read assemblies.
• Systematically evaluated short-read, PacBio HiFi, and Oxford Nanopore data with various callers and alignment strategies.

Main Results:

• Performance varied significantly by inversion class and genomic context; simple inversions were detected with high sensitivity.
• Complex and heterozygous inversions remained challenging to detect accurately.
• Sniffles2 and Severus showed strong recall for complex inversions, though with higher false positives; mapper choice impacted repetitive region detection.

Conclusions:

• This work presents the first unified, high-resolution inversion benchmark.
• It highlights clear strengths and limitations of current inversion detection methods across platforms.
• The resource facilitates principled tool development and evaluation for accurate inversion variation resolution.