The impact of variant annotations on the diagnostic yield of exome sequencing for rare pediatric neurological diseases

Abstract

Variant annotations are crucial for the efficient identification of pathogenic variants. In this study, we retrospectively evaluated the impact of various annotations on identifying 273 pathogenic single nucleotide and small insertion/deletion variants (SNVs/small indels) from 242 patients. Although variant filtering based on allele frequency is essential for narrowing down candidate variants, we found that 13 de novo pathogenic variants in autosomal dominant or X-linked dominant genes had been registered in gnomADv4.0 or 54KJPN with an allele frequency of less than 0.001%, suggesting that very rare variants in large cohort data can be pathogenic de novo variants. Strikingly, 38.1% candidate SNVs/small indels had been registered in the ClinVar database as pathogenic or likely pathogenic, highlighting great utility of this database. SpliceAI can detect candidate variants affecting RNA splicing, leading to the identification of four variants located at 11 to 50-bp away from the exon-intron boundary. Prioritization of candidate genes by patients’ phenotypes using PhenoMatcher module revealed that approximately 95% of the candidate genes had a maximum PhenoMatch score of ≥ 0.6, suggesting the utility of variant prioritization using phenotypes. This study suggests that a combination of multiple annotations and the appropriate evaluation can improve the diagnostic yield of rare diseases.