End-to-end Computational Drug Design for COVID-19: From Screening to Series and Back Again

Austin Clyde, Argonne National Laboratory: COVID-19's rapid emergence required a new drug discovery paradigm leveraging high-performance computing, artificial intelligence, and high-throughput laboratory techniques. At the beginning of the pandemic, the National Virtual Biotechnology Laboratory, a consortium of National Laboratories, began virtually screening compounds in various supercomputing facilities. The result was identifying a molecule with activity against the 3CL-protease of SARS-CoV-2. In this talk, I outline the effort with an emphasis on new computational developments for later stages of the drug discovery process after identifying a lead, particularly novel methods for computational series design and hypothesis testing. Towards thinking about pandemic preparedness and ongoing surveillance drug development, I comment on the scaling dynamics of the current approach to computational drug design and how HPC might adapt its system designs for this task.