Diagnostic utility of clinical genome reanalysis in rare pediatric disorders using long-read sequencing.

Abstract

Over half of presumed genetic disease cases remain undiagnosed following short-read exome (SR-ES) or genome sequencing (SR-GS). Long-read genome sequencing (LR-GS) shows promise for uncovering etiologies missed by SR genetic testing, particularly structural variants (SVs). However, SV interpretation remains challenging due to limitations in call reliability, population allele frequency estimates, and functional impact prediction. To advance clinical LR-GS implementation, we analyzed the genomes of 19 children with suspected rare genetic conditions and prior negative or inconclusive clinical SR-GS/-ES and their parents using PacBio HiFi LR-GS. One additional family with limited DNA underwent Illumina SR-GS only, and 11 probands received SR-GS to complement small variant detection. LR-GS data were processed using phased-assembly and read-based variant calling pipelines validated on SV positive controls, while SR-GS data were processed with the Illumina DRAGEN pipeline. Variants were prioritized using phenotype-driven approaches. Diagnostic variants (likely pathogenic or pathogenic) were identified in 2/20 (10%) families, while an additional 5/20 (25%) harbored findings of uncertain diagnostic significance, including variants of uncertain significance (VUS) and variants in genes of uncertain significance (GUS). All reported variants were detected independently of LR-GS by research SR-GS or by reanalysis of prior clinical SR data. Several LR-GS SV candidates were excluded after population allele frequency filtering, underscoring its importance in clinical SV interpretation. Overall, the observed 10% increase in diagnostic yield was achievable through SR analysis alone, as LR-GS was not required to identify diagnostic variants in this cohort. Functional studies are needed to clarify the clinical relevance of uncertain findings.