The world urgently needs effective drugs for COVID-19. The genome of SARS-CoV-2 encodes about 25 proteins needed by the virus to infect humans and replicate. For example, the spike (S) protein recognizes human ACE2 in the initial stage of infection. Then there are the proteases, which cleave both viral and human proteins, the RNA polymerase, and several others. Drugs acting on these proteins would interfere with essential viral processes, including preventing the synthesis of viral RNA, inhibiting virus replication, blocking the virus binding to human cell receptors or inhibiting the virus’s self-assembly process. Finding a drug that binds to one of these proteins and stops it from working is a rational approach. Structure-based drug discovery (SBDD) was born.
In this process the scientist mimics nature, by using computers to ‘dock’ compounds into binding sites in 3D models of the protein targets. The binding affinity of the compound is calculated, using physics-based equations that quantify the interactions between the drug and its target. The top-ranked compounds are then tested experimentally to see if they indeed do bind and have the required downstream effects on cells and animal models (such as stopping viral infectivity). SBDD has been important in finding antiviral drugs, such as in the 1990s against HIV.
ORNL has unique capabilities in the long and short term that can be applied in the search for small-molecule therapeutics. Nowadays, supercomputers such as Summit at Oak Ridge National Laboratory—the current most powerful in the world—perform massively parallel processing in which many calculations are performed at the same time. This is, perhaps evidently, excellent for docking databases of compounds, but it also can be used in dynamics simulations, where many replicas of the same target can be run in parallel, each exploring a slightly different conformational space. Thus, a comprehensive simulation model of a SARS-CoV-2 drug target protein can be obtained on Summit in a day, whereas it would take months on a typical cluster. Structural biologists are rapidly determining 3D structures for many of the SARS-CoV-2 proteins. The field is thus primed for quick results.
Here, we concentrate on efforts that are likely to be fruitful in the time periods of weeks and months. The de novo development of novel small-molecule therapeutics takes years, with trials and regulatory approval taking between 10 to 15 years in the US (on average). To avoid this, and given that the outbreak of SARS-CoV-2 is a global challenge, it would be of great benefit to identify and repurpose already well-characterized small molecules, such as metabolites, natural products and previously approved drugs, for use in combating the virus.
Our initial aim is therefore to quickly computationally identify drugs for repurposing as hitting SARS-Cov-2 proteins. For this we are making use of the massive power of the Summit supercomputer at ORNL which can minimize time-to-solution for important problems. With this are obtaining the most detailed simulation models possible of the proteins of SARS-CoV-2 and performing virtual high-throughput screening ensemble docking campaigns to find the best-ranked drug candidates. In a parallel proposal they will be tested in vitro for ligand binding.
This project is part of the COVID-19 HPC Consortium.
|Number of Jobs Run To-Date||Node Hours Used To-Date||Number of Jobs Running Now||Nodes in Use Now|