Integration of Variant Annotations Using Deep Set Networks Boosts Rare Variant Association Testing

This talk from Valence Labs' Multiomics Reading Group explores the research paper "Integration of variant annotations using deep set networks boosts rare variant association testing" published in Nature. Learn about DeepRVAT, an innovative model based on set neural networks that addresses the statistical challenges of rare variant analysis in genetic studies. The presentation covers how this approach learns trait-agnostic gene impairment scores from rare variant annotations and phenotypes, enabling both gene discovery and trait prediction. Discover the substantial improvements in gene discoveries and detection of individuals at high genetic risk demonstrated across 34 quantitative and 63 binary traits using whole-exome-sequencing data from UK Biobank. The talk also explains how DeepRVAT enables calibrated and computationally efficient rare variant tests at biobank scale, advancing the discovery of genetic risk factors for human disease traits. Connect with the speakers and access more details by joining the AI for drug discovery community at Portal.