A trio-based long-read sequencing workflow identifies a pathogenic transposable element insertion in a previously undiagnosed patient.

Long-read sequencing (LRS) is a powerful tool for detecting structural variants (SVs), which are major causes of genetic diseases. However, the application of trio-based analysis to LRS data remains challenging due to the complexity of merging and comparing SVs across individuals. Here, we established a workflow to prioritize pathogenic SVs using trio-based LRS whole-genome sequencing. The workflow integrates high-accuracy basecalling, trio-based phasing, and multiple SV callers (CuteSV and Sniffles2). We employed the PanPop Realign and Thin (PART) process to merge SVs and utilized the JSV1 Japanese population-specific SV frequency dataset to filter common variants. We evaluated this workflow using 12 family trios from the Initiatives for Rare and Undiagnosed Diseases (IRUD) project. The workflow successfully prioritized candidate SVs by narrowing down tens of thousands of loci to less than ~20 with X-linked homozygous and de novo models for manual inspection. Notably, we identified a pathogenic AluY element insertion in the GPC3 gene in a patient with Simpson-Golabi-Behmel syndrome. This variant had been missed by previous short-read whole-exome sequencing but was clearly prioritized by our LRS workflow. Our workflow effectively prioritizes pathogenic SVs from LRS data, demonstrating the clinical utility of LRS for diagnosing rare diseases involving complex SVs without prior knowledge of candidate genes or loci.