Identification of de novo variants from parent-proband duos via long-read sequencing

While de novo variants cause many Mendelian disorders, their detection currently requires sequencing of the proband and both biological parents. This is not feasible when only one parent is available, a limitation for millions of families. Here, we develop duoNovo, which identifies de novo variants from parent-proband duos using long-read sequencing followed by haplotype reconstruction and detection of identical-by-descent haplotype blocks. We sequenced 104 trios with PacBio HiFi sequencing and applied duoNovo to each of the 208 duos constructed by masking one parent, classifying over 55 million variants according to their de novo status. We evaluated duoNovo's performance against classifications obtained using the full trios (which included over 10,000 de novo variants), finding a positive predictive value of ∼98% among variants absent from gnomAD and a sensitivity of approximately 55% from father-proband duos (73% of the biological upper limit) and 14% from mother-proband duos (56% of the biological upper limit), the latter increasing to ∼24% when a sibling is available. In a cohort of 63 undiagnosed duos, duoNovo provided critical diagnostic information for two probands. In summary, duoNovo has the potential to significantly increase the diagnostic yield of single-parent genetic testing and represents an example where long-read sequencing provides a clear benefit over short-read sequencing even for single-nucleotide variants. It is freely available as an R package.