Detecting pathogenic structural variation in families with undiagnosed rare disease in a national genome project.

Abstract

Whole-genome sequencing (WGS) projects for rare disease diagnosis typically yield a diagnostic rate of 25-41%, depending on the methods for patient selection and the extent of prior genetic testing. The Scottish Genomes Partnership (SGP) is a collaborative programme using genome sequencing to diagnose rare disease patients with presumed monogenic aetiology in the Scottish NHS. Within SGP, short-read sequencing (SRS) had previously achieved a diagnostic rate of 23% in affected families. To increase diagnostic yield, we applied Oxford Nanopore Technologies (ONT) long-read sequencing (LRS) to a cohort of 24 SGP families (74 individuals) that remained undiagnosed after SRS-based SNV/indel analysis. We also retrospectively reviewed SRS-derived structural variant (SV) calls to assess whether LRS results could have been detected in SRS data. After quality, rarity, panel-based and inheritance-based filtering, 392 candidate SVs were retained across de novo, homozygous recessive, compound heterozygous and X-linked inheritance models, of which 8 overlapped PanelApp diagnostic-grade "green" genes. Pathogenic or likely pathogenic de novo SVs were identified in 3 of 24 families: an AUTS2 inversion, a DLX5/6 locus inversion, and an FN1 deletion. All three SVs were independently confirmed by retrospective SRS re-analysis using SVRare. These findings demonstrate that LRS-based SV analysis, supported by orthogonal SRS re-analysis, can resolve clinically significant SVs in families who remain unsolved after standard rare disease testing.