- High-precision 3D simulations reveal how SARS-CoV-2 virus attaches to human cells
- Detailed understanding of the spike protein interaction needed to evaluate drug candidates
- Advanced molecular dynamics simulations on supercomputers will give higher accuracy results
A virus may not technically be alive—but it sure can do a lot of damage to living things. Viruses are extremely small organisms that consist of molecules of DNA or RNA surrounded by a protective protein coat.
Despite possessing genetic material, viruses reproduce not through cell division but via spontaneous assembly within a host cell. In the case of SARS-CoV-2, that host cell is human.
The virus that causes COVID-19 is covered in protein spikes that attach to the ACE2 human cell receptor. That’s how a person becomes infected with the virus. Many scientists think that better understanding how the spike protein attaches to ACE2 will be the key to lessening its devastating impact on human health.
New research led by theoretical chemist Jean-Philip Piquemal of Sorbonne University in Paris, will use molecular dynamics (MD) simulations and high-performance computing (HPC) to do just that.
“Our study aims to model the interaction of the virus Spike surface protein and the ACE2 human cell receptor,” says Piquemal. “Inhibiting the Spike recognition with the ACE2 cell receptor would neutralize the virus by preventing it from entering the cell.”
Other components of the virus are also ripe targets for intervention. For example, if SARS-CoV-2 can’t retain the structure of its functional proteins, it would almost entirely lose its ability to replicate, significantly reducing the threat to human hosts.
Our plan is to attack COVID-19 from various angles to be able to fight it as effectively as possible ~Jean-Philip Piquemal
But when you’re talking about an organism that is 1/100th the size of the average microscopic bacteria, that’s easier said than done. That’s why scientists use molecular dynamic simulations to gain insight into the behavior of viruses at the atomic level.
Thanks to an emergency grant from the Partnership for Advanced Computing in Europe (PRACE), Piquemal and his team will use supercomputers to perform a detailed, all-atoms simulation of the protein machinery of the SARS-Cov-2 virus.
“Tackling COVID-19 research with advanced molecular dynamics is a real HPC challenge,” says Piquemal. “The PRACE Fast-Track process was quick and within one week offered us the possibility of using the GENCI Irène Joliot Curie machine located at TGCC in France in order to get access to the latest available CPUs and GPUs."
His team’s molecular dynamics code, Tinker-HP, will incorporate high-precision polarizable force fields into their model of the SARS-CoV-2 virion. They hope this added complexity will lead to higher accuracy results when they simulate interactions of potential new drugs with the 3D model of the virus.
Supercomputers enable us to build and simulate high-precision 3D models of the different proteins of the virus ~Piquemal
“These models can be used to better understand the structural aspects of the SARS-CoV-2 virus," says Piquemal. We can see all of the details—the same ions, the same atoms, the same molecules—as closely as we can from experiments.”
Piquemal and his team have actually been working on the Tinker-HP code for over ten years. It was designed to perform biomolecular simulations of viruses, but until recently it has mostly been used to study HIV.
Transitioning to a focus on coronavirus means the team has had to make big changes incredibly quickly. The existing code was already running on CPUs, and they had plans to scale up for use on GPUs in the future. But the coronavirus outbreak and the urgent need for more information about the pathogen means they have had to speed up that timeline.
“In the last eight weeks we have been working full-time on getting the code ready for GPUs,” says Piquemal. “It was nearly there, but wasn’t intended to be published or tested by anyone before September or October.”
We really want to get to the highest level of accuracy and detail in simulating the virus ~ Piquemal
He continues, “What’s missing now is that nobody has an idea of the full virus structure. The science is not there, even if the code was able to do it. We need more details from experimentalists—It’s a global problem of understanding how this virus works.”
Fortunately scientists like Piquemal have a lot of experience in global collaboration. Tinker-HP was developed in collaboration with researchers such as Pengyu Ren at the University of Texas at Austin and XSEDE, among others. And they will continue to work with researchers around the world to make new discoveries in the short time the PRACE project allows.
“In six months, we hope to have a very detailed understanding of the spike protein and to compute the strength of the interaction between the spike protein and the receptors from the human cell,” says Piquemal. “We want to make available high-level, high-resolution simulations that other people can pick up.”
Matthieu Montes of CNAM will lead efforts at drug discovery based on the simulation results, but the simulations will also be released to the wider scientific community to make their own discoveries.
Piquemal is excited to be part of the collaboration. He thinks the urgency of the COVID-19 crisis has the potential to accelerate knowledge breakthroughs. He says, “The dangers of our current situation are creating the moment for a technological jump.”