Assessment of the variant prioritisation strategy for genomic newborn screening in the Generation Study

Purpose Genomic sequencing offers the opportunity to screen for hundreds of rare genetic conditions with a single test. To minimise potential negative impact on families and clinical services, it is crucial to reduce false positive results while prioritising clinical utility. Here we present an automated variant prioritisation approach for genomic newborn screening and the assessment of its validity across the conditions included in the Generation Study, a research study investigating genomic sequencing in 100,000 newborns in England. Prioritised variants will subsequently undergo manual review by a registered clinical scientist, and a specialist clinician, before being reported back to parents. Methods The automated variant prioritisation approach was evaluated for its ability to predict the presence or absence of relevant genetic variants, subsequently informing gene and variant-specific inclusion in the study. Specificity was estimated using a cohort of 34,410 samples not enriched for rare diseases, while sensitivity was assessed with 546 samples from participants with diagnostic variants in genes relevant to newborn screening. Coverage and CNV callability metrics were evaluated as proxies for variant detection. Results Assessing validity of automated variant prioritisation on a gene level led to changes in rules used in variant prioritisation and conditions included. We estimated that 3-5% of samples will have prioritised variants that require manual review and that <1% of samples will be taken forward for orthogonal confirmatory testing and genetic confirmation. We prioritised variants in ~80% of samples with diagnostic variants across genes included in the Generation Study. Conclusion Gene-specific assessment of variant prioritisation is crucial to establish analytical validity prior to inclusion in genomic newborn screening.