Leveraging High Performance Computing and AI to Advance Biomedical Discovery

In 2016, The Department of Veterans Affairs (VA) and the Department of Energy (DOE) have established an Interagency Agreement (IAA) to combine VA's vast array of clinical and genomic data with DOE's national computing capabilities-including the most powerful supercomputer in the nation-to push the frontiers of precision medicine to support the President's National Precision Medicine Initiative (PMI) and improve the lives and wellbeing of the nation's veterans. Guided by the vision to improve the health and wellbeing of the nation's veterans and the general public, the MVP Champion (Computational Health Analytics for Medical Precision to Improve Outcomes Now) program will develop better methods to treat, cure, detect at early stages, and prevent diseases through precision medicine by harnessing the genomic and clinical data in the VA's Million Veteran Program (MVP) and the computational power within the DOE. In this talk, we will present results from this project that led to creation of new computational methods that scaled to both the genetics data and computational capabilities that are made available for this project. We will especially, focus on the genome-wide PheWAS study, where we performed 300 billion association studies on more than half of DOE's supercomputer called Summit, accelerating the analysis pipeline 100 fold over current state of the art resulting in discovery of over 38K novel biomarkers [1]. Additionally, we will present ongoing work on developing AI/ML technologies for disease risk prediction, diagnosis and treatment.