AI for Health by Artefact

🇬🇧 Janssen session - Leveraging large-scale population-based genomics data to inform drug discovery and development

Nov 17, 2021 | 4:48 PM CET - 5:00 PM CET


Over the past two decades, only ~10% of drugs that entered phase 1 clinical trials were successfully approved by FDA, and roughly half of those that failed were due to non-clinical toxicology or clinical safety concerns. On the premise that phenotypes arising from variation in a gene encoding a protein can be used to predict those that would result from drug modulation of that protein, we developed a workflow that leverages the large-scale genomic data available in the UK Biobank and ancillary datasets available in the public domain, to predict adverse events that may be observed in the clinic. An analytical pipeline for the implementation of our systematic approach to a broad spectrum of targets, including our ongoing development of a machine learning framework that allows for estimation of a target-specific liability index, will be described. Lastly, a contemporary use case for prediction of adverse outcomes resulting from inhibition of LRRK2 in the treatment of Parkinson’s disease will be presented.