Expanding drug targets for 112 chronic diseases using a machine learning-assisted genetic priority score

• DOI: 10.1038/s41467-024-53333-y
• Publication Date: 2024-10-15T14:43:21Z
• Authors: Robert Chen; Áine Duffy; Ben O. Petrazzini; Ha My Vy; David Stein; Matthew Mort; Joshua K. Park; Avner Schlessinger; Yuval Itan; David N. Cooper; Daniel M. Jordan; Ghislain Rocheleau; Ron Do
• Keywords: Drug Development

Identifying genetic drivers of chronic diseases is necessary for drug discovery. Here, we develop a machine learning-assisted priority score, which call ML-GPS, that incorporates associations with predicted disease phenotypes to enhance target First, construct gradient boosting models predict 112 phecodes in the UK Biobank and analyze observed common, rare, ultra-rare variants model allelic series. We integrate these existing evidence using continuous feature encoding training it indications Open Targets externally testing SIDER. then generate ML-GPS predictions 2,362,636 gene-phecode pairs. find use phenotypes, identify substantially more than across allele frequency spectrum, significantly improves performance ML-GPS. increases coverage targets, top 1% all scores providing support 15,077 pairs previously had no support. can also well-known target-disease relationships, promising targets without indicated drugs, several drugs clinical trials, including LRRK2 inhibitors Parkinson's olpasiran cardiovascular disease.