NERSCPowering Scientific Discovery Since 1974

COVID-19 Support at NERSC

NERSC is supporting COVID-19 related research by providing urgent access to its High Performance Computing and Data resources and staff expertise. All interested researchers can apply to use NERSC  through the COVID-19 High Performance Computing Consortium. This unique private-public effort brings together federal government, industry, and academic leaders to volunteer free computing time and resources on their world-class machines. Scientists are invited to submit COVID-19 related research proposals. An expert panel of top scientists and computing researchers will select projects based on public health benefits, with an emphasis on rapid results. Proposals are reviewed and awards matched to resources with a very fast turnaround (one to two days). 

NERSC has initially set aside up to 1.25 million node hours on its Cori supercomputer in support of this effort. Personalized staff assistance and fast-turnaround queues are also available for COVID-19 research.

NERSC is currently supporting the following projects performing research related to COVID-19. In addition to working with projects awarded from the COVID-19 HPC Consortium, NERSC is  collaborating with a number of other research teams as described below.

Drug-Repurposing for Covid-19 with 3D-Aware Machine Learning

Award from the COVID-19 HPC Consortium
Investigators: Rafael Gomez-Bombarelli (Massachusetts Institute of Technology), Simon Axelrod (Harvard University)

Rafael Gomez-Bombarelli and Simon Axelrod

Rafael Gomez-Bombarelli (left), is the Toyota Assistant Professor in Materials Processing at MIT. Simon Axelrod is a graduate student in Chemical Physics at Harvard University

Novel active therapeutics against coronaviruses like the one responsible for Covid-19 (SARS- CoV2) are in urgent need. New discovery of small-molecule drugs, however, is slow and costly. Many promising molecules that strongly interact with their biological target fail as drugs because of poor processing within the body or toxicity that is only discovered late in their development. Furthermore, the synthesis-experimentation loop is typically slow. Repurposing known drugs for the treatment of a new disease, such as Covid-19, effectively bypasses many of these challenges, since the drugs are already FDA approved.

In this project, the team is exploring whether the repurposing of drugs for Covid-19 treatment can be accelerated with a combination of physical simulation and machine learning (ML). They are using affordable electronic structure simulations to calculate molecular conformations and train 3D-based message-passing neural networks from existing molecular screens against the related SARS-CoV1 and SARS-CoV2 data as it becomes available.

Publications: "GEOM: Energy-annotated molecular conformations for property prediction and molecular generation"

Computer-Aided Design of Peptide Ligands to SARS-CoV-2 Targets

Investigators: Paul Keim, Nagarajan Vaidehi, Supriyo Bhattacharya, John Altin, Erik Settles, Jason Ladner
Affiliation: Beckman Research Institute of the City of Hope
More Information: NERSC’s Scientific Computing Power Steps into Coronavirus Battle

Supriyo Bhattacharya

Supriyo Bhattacharya is an Assistant Research Professor at the City of Hope where he works on small molecule drug design, peptide and antibody design, protein-protein interaction inhibition, RNA design, nanoparticle design for drug delivery, understanding protein conformational dynamics.

The goal of this project is to design peptides and small molecules that will bind to the coronavirus surface proteins and inhibit their binding to human proteins. Besides potential therapy, these inhibitory peptides and molecules are useful for understanding the mechanisms of virus entry and interaction with the human host and the immune system.

The project will start by simulating the viral Spike protein bound to human ACE2 which the virus uses for invading the lungs. NERSC computing resources will be used for running MD simulations on the complex of Spike bound to ACE2. These simulations will be then used for designing peptides and small molecules to disrupt the interactions between Spike and ACE2. The team has a collaboration with collaborators at TGen and Northeastern University to test the binding of the designed peptides and compounds.

Small molecules can be used as probes for understanding the interaction of the SARS-Cov2 virus with its host and offer future potential as diagnostic and therapeutic agents. Protein-protein interactions are critical to viruses as they infect host cells and the ability to disrupt this with small-molecule ligands will lead to a greater understanding of viral biology. The most prominent feature of the coronavirus virion is the spike protein and it is known to be critical in the infectious process. We will use the molecular structure of the spike protein to identify potential ligand binding sites and then design peptides (11-20 amino acids) as potentially high-affinity ligands. Our computational search will identify a large number of potential candidates that can be included in a single peptide library and tested empirically for binding. A computational search for peptide ligands that bind to the spike protein will be performed using the Rosetta algorithm in conjunction with distributed high performance computing. The best peptide sequences will then be connected together by peptide linkers to attain further affinity gain. The top 100,000 peptide sequences obtained by this approach will be synthesized for rapid testing in in-vitro assays for Spike binding. The top peptide candidates will also be tested on SARS-Cov-2 grown on human tissue culture cells and murine COVID19 models that are being developed.

Video: One microsecond molecular dynamics simulation of SARS-Cov2 Spike bound to human ACE2; The Spike is colored in cyan and ACE2 in orange. The surface glycans are shown as spheres.

Structure Refinement of the Spike Protein of SARS-COV-2

Ching Wai Yum

Wai-Yim Ching is a Professor in the Dept. of Physics and Astronomy at the U. of Missouri-Kansas City.

Investigator: Wai-Yim Ching
Affiliation: University of Missouri-Kansas City

Award from the COVID-19 HPC Consortium

The current COVID-19 pandemic is a severe threat in human health leading to social, and economic disruption in the entire world. The spike (S) glycoprotein of SARS-COV-2 is considered to be a key target for vaccine development since it makes first contact to ACE2 acceptor in human protein. The structure of the spike protein was recently determined by cryo-EM techniques with a resolution of 3.5 Å. It contains three chains and each chain consists of four structural domains: receptor binding domain (RBD), n-terminal domain (NTD), subdomain 1 and 2 (SD1-SD2) and S2. This project will refine the structures of these domains to much higher accuracy using large scale ab initio simulations and investigate their atomic scale interaction and intra molecular binding mechanism. They will also present electronic structure calculations with detailed intra and intermolecular interactions of the spike protein in the pre- and post-fusion configuration which are difficult to determine experimentally. In addition, they are studying the further interaction between the spike protein and ACE2 to shed some light on the genetic modification due to mutation and thereby provide useful insights for various vaccine development and drug design.

Publications: Intra- and intermolecular atomic-scale interactions in the receptor binding domain of SARS-CoV-2 spike protein: implication for ACE2 receptor binding, Adhikari, Puja and Li, Neng and Shin, Matthew and Steinmetz, Nicole F. and Twarock, Reidun and Podgornik, Rudolf and Ching, Wai-Yim. Phys. Chem. Chem. Phys., 2020, DOI: 10.1039/D0CP03145C

COVID Scholar

COVID Scholar team photo

Berkeley Lab researchers (clockwise from top left) Kristin Persson, John Dagdelen, Gerbrand Ceder, and Amalie Trewartha led the development of COVIDScholar.  (Credit: Berkeley Lab)

Investigators: John Dagdelen, Kristin Persson, Gerbrand Ceder, Amalie Trewartha
Affiliation: Lawrence Berkeley National Laboratory
More Information Machine Learning Tool Could Provide Unexpected Scientific Insights into COVID-19


A team of materials scientists at Lawrence Berkeley National Laboratory (Berkeley Lab) – scientists who normally spend their time researching things like high-performance materials for thermoelectrics or battery cathodes – have built a text-mining tool in record time to help the global scientific community synthesize the mountain of scientific literature on COVID-19 being generated every day.

This project is text miningCOVID-19 publications for new scientific insights using unsupervised textual analysis described in the Nature paper, “Unsupervised word embeddings capture latent knowledge from materials science literature.” All the project's infrastructure is running at NERSC.

Dependence of Structure and Dynamics of Novel SARS-CoV-2 on Temperature and Humidity in the Atmosphere

Soumya Rath

Soumya Rath studies proteins in diseases at the National Inst. of Technology Warangal, where she is an Asst. Professor in the Dept. of Biotechnology

Award from the COVID-19 HPC Consortium
Investigators: Soumya Rath, Kishant Kumar
Affiliation: National Institute of Technology Warangal

The havoc caused by the novel SARS-CoV-2 coronavirus has spread across almost all the nations of the world. The pandemic took place due to the highly contagious nature of the virus which can be easily transmitted from humans to humans. Others in the coronavirus family of viruses such as SARS and MERS have infected individuals. However, they gradually disappeared in  hot and humid conditions. Data obtained from countries with higher temperatures and humidity also indicate a low propensity of infection. This work studies whether the virus undergoes any biophysical changes in response to changes in atmospheric conditions, mainly temperature and humidity, on the virus to determine if it undergoes any biophysical changes with change in atmospheric conditions. The study uses molecular dynamics (MD) simulations that solve Newton’s law of motion and deterministically determine the position and momentum of atoms in a system. The virus structural proteins will be modeled first in atomistic and later in coarse-grained (CG) methods and then the difference in dynamics would be analyzed. This study would open a new dimension in the characterization of SARS-CoV-2 and future corona family class of viruses in prevention, categorization, and drug designing aspects.

Publications: "Investigation of the effect of temperature on the structure of SARS-Cov-2 Spike Protein by Molecular Dynamics Simulations"

Manipulation of Membrane Curvature by Designer Peptides for Battling COVID-19 Infection 


Qiang Cui, a Professor of Chemistry at Boston University, studies a diverse set of chemical and biological problems.

Award from the COVID-19 HPC Consortium
Investigator: Qiang Cui
Affiliation: Boston University

The G. C. L. Wong lab at UCLA recently identified antimicrobial peptides (AMP)-like sequences in the SARS-CoV-2 genome with high membrane remodeling activity, a process that affects key events in the infection cycle of the virus. In preliminary studies, they further showed evidence that creation of new "inverse translocation peptides" (ITPs) could lead to a potentially effective strategy that mitigates COVID-19 infection. The goal of this project’s computational study is to work closely with the Wong lab to understand the factors (sequence and membrane composition) that dictate the membrane activity of ITPs and their effect on the structure of the membranes by the AMP-like sequences of SARS-CoV-2. Insights from these studies will guide the design of new ITPs for battling virus infection. The computations will combine atomistic and coarse-grained simulations.





Stanislav Stoyanov is a computational chemist with extensive and considerable experience in quantum chemistry, molecular dynamics, statistical mechanics and methods for coupling these.

The Competition of Antiviral Drugs with ATP to Inhibit the SARS-CoV-2 RNA-dependent RNA Polymerase: A Key to Enhanced Drug Screening 

Award from the COVID-19 HPC Consortium
Stanislav Stoyanov, Sergey Gusarov
University of Alberta, Natural Resources Canada, National Research Council Canada

This project aims to demonstrate a novel computational approach for the enhanced screening of drugs, considering their effects in solution with ATP. The project addresses the question of why molecular dynamics studies of some drugs’ behavior in isolation show strong interference with viral replication, but clinical trials show lower impacts than might be expected.



Simulations of Mutated Indinavir and Hydroxychloroquine-SARS-CoV-2 Protease Complexes

Award from the COVID-19 HPC Consortium
Investigators: Jithin Sunny
Affiliation: Shri Ramaswami Memorial University of Science and Technology

Much effort has been made to understand the drug-protein interactions for developing anti-viral therapies. To target viral infections, analyzing potential inhibitors targeting their proteins has been a widely used approach. The current SARS-CoV-2 protease is a promising target and has been used with a combination of many anti-protease drugs to inhibit its function. Our initial analysis has shown Indinavir and Hydroxychloroquine to bind with protease with a higher affinity than the currently studied drugs. However, rapid evolution in viral genomes may sometimes lead to the acquisition of mutation in proteins, leading to variation in drug binding. Thus, besides the initial docking studies, we also evaluated the binding affinity of the two drugs with several mutant protease structures. Preliminary results have shown Hydroxychloroquine to be effective against mutating SARS-CoV-2 protease. With an aim to further validate these results, we wish to perform molecular dynamics simulations. MD simulations are necessary to demonstrate the binding stability of the drug-protein complex. The stability of the mutant Hydroxychloroquine-protease complexes over the time period of 50-100 nanoseconds will determine whether the drug is indeed responsive towards a mutating SARS-CoV-2 protease. Using the computational resources provided by the consortium, the GROMACS tool can be used to study the changes in significantly less time. 


Health and Human Services (HHS) Epidemiology Studies

Investigators: Diana Wong, Nick Robison
Affiliation: U.S. Department of Health and Human Services

This team is using GCM, an agent-based epidemiological model being used by FEMA, to inform policymakers about the number of COVID-19 cases that are projected to occur in various regions, the need for medical resources, and to understand the impact of social distancing and other interventions. The code is written in Java using shared memory parallelism, and while individual states can be modeled on standard computing nodes, a model of the entire US, which includes interactions between states, requires very large memory. NERSC worked with the GCM team to run a COVID19 simulation for the full U.S. population on a terabyte memory system, which will be used to compare and validate the stay-by-state models that are run frequently for different scenarios. Additional full U.S. simulations are planned at NERSC as the situation evolves.

Large Scale Docking for COVID-19

Investigators: John Irwin
Affiliation: University of California, San Francisco

The aim of this project is to build 3D databases of commercially available molecules and dock them into the binding site of COVID-19 target proteins. The database will be ranked best-to-worst, and top-scoring molecules will be prioritized for purchase and experimental testing. This method allows us to consult thousands of times more chemistry for ligand discovery that is possible using traditional methods.


Investigators: Peter Nugent, Peter Harrington, Nicholas Choma, Jan Balewski
Affiliation: Lawrence Berkeley National Lab

ExaLearn is a project within DOE's Exascale Computing Project

This team is developing a deep learning-based surrogate model for Epidemiology codes. They are running FluTE, a publicly available stochastic influenza epidemic simulation model, and Corvid, an agent-based SARS-CoV-2 (COVID 2019) transmission model, on Cori.