Making sense of RNA splicing variants

Abstract

Determining clinically whether a coding or non-coding variant will actually disrupt pre-mRNA splicing and cause a genetic disorder is one of the grand challenges currently facing genetic pathology. Variant curation is a time-consuming use of highly skilled labour and the rate-limiting step for laboratory throughput. Splicing variant interpretation compounds these issues, exacerbated by the exponential growth of genomic testing. This talk will highlight advances in machine-learning and evidence-based approaches to predict if a DNA will disrupt mRNA splicing, and if so, how – exactly? Leveraging confirmed diagnostic and clinical outcomes for ∼100 families triaged in real-time into research-led RNA diagnostic testing, we will discuss: strengths and weaknesses of different modes of RNA diagnostic testing; retrospective evaluation of the predictive accuracy of artificial intelligence splicing prediction tools; and, the clinical utility of empirical methods to reliably predict the nature of mis-splicing for pathology use of PVS1 or PP3/BP4, and/or, to direct strategic design of RNA Diagnostic assays.