Abstract
Genetic studies of Mendelian and rare diseases face the critical challenges of identifying pathogenic gene variants and their modes-of-action. Previous efforts rarely utilized the tissue-selective manifestation of these diseases for their elucidation. Here we introduce an interpretable machine learning (ML) platform that utilizes heterogeneous large-scale tissue-aware datasets of human genes, and rigorously, quantitatively and concurrently assesses hundreds of candidate mechanisms per gene. Application of our tissue-aware ML platform to Mendelian disease genes enabled to pinpoint known and previously-underappreciated factors that contributed to disease manifestations. Next, we extended our ML platform toward genetic diagnosis of rare diseases with tissue-selective manifestations. When applied to genetic data from 50 patients, patient-specific ML models successfully prioritized the pathogenic gene. Lastly, analyses of tissue-selective complex traits resulted in trends resembling those observed for Mendelian diseases. Thus, interpretable tissue-aware ML models can boost genetic diagnosis and mechanistic understanding of tissue-selective heritable diseases and complex traits.
Zoom connection information
Join Zoom Meeting
https://uio.zoom.us/j/63276550975?pwd=OGJvWGVDbUxWdHRXeWczbGxxZEJpZz09
Meeting ID: 632 7655 0975
Passcode: 359444
Junior talk
Tatiana Belova, Postdoctoral fellow at the Centre for Molecular Medicine Norway (NCMM), UiO, will present her work about "Capturing cancer heterogeneity using patient-specific gene regulatory networks."
.