Abstract
BACKGROUND: Singleton short-read genome sequencing (GS) is increasingly used as a first-line genetic test for childhood neurological disorders (such as intellectual disability, neurodevelopmental delay, motor delay, and hypotonia) with diagnostic yields from 26 to 35%, typically involving a mix of single nucleotide variants and small insertions/deletions (SNV/INDELs), structural variants (SVs), and short tandem repeats (STRs). Long-read GS is emerging as an attractive alternative, offering a more comprehensive assessment of the genome, but its utility still needs to be systematically evaluated in a clinical diagnostic setting. METHODS: We prospectively included 100 children and adolescents (≤ 20 years) with neurological disorders, newly referred for genetic testing. Routine DNA was used for singleton standard clinical short-read GS in parallel with long-read GS (Oxford Nanopore Technologies). In addition to comprehensive variant calling, long-read GS data was also phased and underwent methylation analysis. Variant interpretation was restricted to in-silico gene panels targeting either intellectual disability (1,568 genes) or neuromuscular disorders (1,035 genes) depending on the clinical presentation. RESULTS: The long-read GS generated an average of 111 GB data per sample, with a median read-length of 5 kb and average N50 of 16 kb; resulting in an average coverage of 34X. Short-read and long-read GS identified the same 29% diagnostic yield, including SNV/INDELs (n = 18), SVs (n = 9), STRs (n = 1), and uniparental disomy (n = 1). Long-read GS provided additional diagnostic value in 13 cases involving 17 distinct variants, including phasing of SMN1 and biallelic SNVs/INDELs in autosomal recessive genes, accurate determination of STR length and sequence as well as detailed structural characterization of SVs. Of note, an unbalanced translocation, der(14)t(8;14)(p11.2;p23.1), required de novo assembly and T2T-CHM13 alignment to resolve the breakpoint junctions. Furthermore, long-read GS detected disease-associated aberrant methylation patterns in the Prader-Willi region and across an FMR1 expansion. CONCLUSIONS: In a clinical diagnostic setting, long-read GS proved to be a streamlined, first-line test, capturing the full spectrum of disease-causing variants, reducing the need for follow-up testing and enabling more precise interpretation. While the overall diagnostic yield may be comparable to that of short-read approaches, long-read GS offers significant added value across multiple variant types.