Pandemic Drugs at Pandemic Speed: Infrastructure for Accelerating\n COVID-19 Drug Discovery with Hybrid Machine Learning- and Physics-based\n Simulations on High Performance Computers

Abstract

The race to meet the challenges of the global pandemic has served as a\nreminder that the existing drug discovery process is expensive, inefficient and\nslow. There is a major bottleneck screening the vast number of potential small\nmolecules to shortlist lead compounds for antiviral drug development. New\nopportunities to accelerate drug discovery lie at the interface between machine\nlearning methods, in this case developed for linear accelerators, and\nphysics-based methods. The two in silico methods, each have their own\nadvantages and limitations which, interestingly, complement each other. Here,\nwe present an innovative infrastructural development that combines both\napproaches to accelerate drug discovery. The scale of the potential resulting\nworkflow is such that it is dependent on supercomputing to achieve extremely\nhigh throughput. We have demonstrated the viability of this workflow for the\nstudy of inhibitors for four COVID-19 target proteins and our ability to\nperform the required large-scale calculations to identify lead antiviral\ncompounds through repurposing on a variety of supercomputers.\n