The world’s battle against the invisible enemy known as the coronavirus continues on all fronts with HPC making a significant contribution. PRACE has now awarded thirty projects under its Fast Track Call for Proposals to support the mitigation of the impact of the pandemic.
The ten projects referenced in this article will receive 91 712 000 core hours and 350 000 node hours on various world-class systems around Europe.
While social distancing and safety regulations are becoming fixed features of our daily lives, more and more people are wondering why the war against the coronavirus hasn’t been won yet. Is this battle really so incredibly complicated? Doesn’t science have enough modern and smart tools against an enemy, as small as 0.1 microns in diameter? Actually, the combat is far trickier than we could imagine. We must never forget that viruses originated long before Homo sapiens: they appeared around three billion years before our ancestors. According to some theories, viruses existed before any other types of life and may have played a major role in shaping life as we know it.
There are studies and hypotheses that posit that somewhere during this period the viruses infected the three main domains of life: bacteria, archaea, and eukaryotes, which began separating at least 3 billion years ago during the Archaean eon. In other words: viruses are as old as life itself.
Unlike viruses, Homo sapiens appeared on Earth only 200 000 years ago. Despite being a relatively “young” form of life, the complex human body has adapted and evolved over the centuries, and the human immune system is capable of defending itself against all kinds of microscopic threats including many viruses. Where in past ages humanity would pay the an extremely high price for immunity – namely the deaths of those unable to withstand the infection – our unique ability to think abstractly has allowed us to come up with counterattacks such as vaccines and hygiene measures.
Today, abstract thinking in combination with the unique computational tools for studying living nature at the atomic and molecular level are the weapons we use to find the coronavirus’ weaknesses and devise ways to neutralise it.
So far, according to various experts from around the world, more than 320 million substances have been tested by computer simulations to determine whether any of them could counteract the virus. Unfortunately, none of them can yet successfully block SARS-CoV-2 in the human body.
PRACE provides access some of the world’s most powerful supercomputers to scientists to gain new knowledge about the virus as quickly as possible. To this end, PRACE awarded ten additional projects which aim to provide even better knowledge of the coronavirus, its prevention, and how to stop it from entering human cells. Some of the tactics used by this batch off projects are:
- Tracking infected droplets through ambient air after a sneeze or cough to set the ideal safety distance between people
- Testing known drugs such as heparin for their possible effectiveness against SARS-CoV-2 in human cells
- Revealing the replication machinery of the coronavirus
With the accumulation of more and more knowledge about this virus, the powerful supercomputers PRACE provides are crucial to achieve breakthroughs.
If you are curious about these projects and results they will yield, here are their summaries:
The project COVID-DROPLETS is led by Dr Gaetano Sardina from Chalmers University of Technology, Sweden, and aims to investigate the lifetime of expiratory droplets released by an individual infected with SARS-CoV-2 (coronavirus strain causing COVID-19). Surprisingly, the current recommendations to hinder the transmission of respiratory infectious diseases are based on a simple model developed 90 years ago.
The assumptions of this model rely on the observation that large droplets (> 10 mm) tend to settle while smaller droplets evaporate faster than they fall in a fraction of second. But the team of Dr Sardina assumes that the scenario is more complicated and to prove that, the scientists will model sneezes and coughs as turbulent jets. Additionally, they will model speaking and breathing via homogeneous isotropic turbulence. They will simulate different levels of atmospheric temperature and humidity to detect the most favourable weather conditions that can enhance or hinder evaporation and transmission of the disease.
The researchers also want to test whether there is a link between air pollution and the spread of the virus. To this end, the scientists will check different levels of PM concentration and calculate the collisions with the pathogenic droplets to assess the hypothesis of a potential connection with the growth of COVID transmission.
The final goal of the project will be guidelines for policymakers to slow down the spread of the pandemic. These guidelines would help them to set a more realistic safe distance between individuals, regulate temperature/humidity of internal public environments to accelerate the evaporation of the pathogen-bearing droplets, and order targeted lockdowns when particular weather or/and high-pollution events occur.
According to the team, these guidelines could be useful for any epidemic transmissions associated with respiratory pathogens. To perform these complicated calculations and simulations, PRACE awarded the project with 20 000 000 core hours on Joliot-Curie Rome, hosted by GENCI at CEA, France.
The project entitled “DyCoVin – Interactions and dynamics of SARS-CoV 2 spike-heparin complex”, is led by Prof. Rebecca Wade from Heidelberg University, Germany.
The spike glycoprotein 1 of SARS-CoV-2 (SG1-Cov2) mediates the binding of the coronavirus to human cell receptor angiotensin-converting enzyme 2 (ACE2). Along with this, heparan sulphate proteoglycans (HSPGs) are used as the first attachment site, leading to an increased concentration of viral particles on the target cell surface. Moreover, in vitro assays suggest that heparin can effectively prevent the coronavirus strain HSR1 from binding. The curious thing, in this case, is that heparin is a very old drug, discovered in 1916.It is an anticoagulant, preventing the formation of blood clots and growth of existing clots. It is used in the treatment of lung diseases such as pulmonary fibrosis, bronchial asthma and asthma-induced airway hypersensitivity, and recent clinical trials suggest that inhaled heparin for lung diseases is beneficial and safe.
The objective of the project team is to pinpoint the role of HSPGs in spike SG1-Cov2 infection using realistic computer simulations. This knowledge will allow characterising the structure and dynamics of putative binding patches for heparin-like compounds on the spike receptor, which needs to adopt the open form to bind ACE2. For that reason, the team aims to stabilise the receptor using different heparin chains to shield binding sites and to decrease the flexibility of the spike.
The researchers expect the impact of this approach for treating the viral infection to be high because the FDA already approved the heparin. That is why aerosol drug administration could provide the advantage of directly delivering heparin to the site of the SARS-CoV-2 infection and thereby to stop the interaction between the virus and the receptor.
PRACE awarded the project with 3 520 000 core hours on Marconi100, hosted by CINECA, Italy.
Exploring COVID-19 Infectious Mechanisms and Host Selection Process is led by Prof. Modesto Orozco from the Institute for Research in Biomedicine (IRB Barcelona), Spain.
The main goal of Orozco’s team is to understand the evolutionary path driving the virus from a bat to humans, to forecast human sensitivity to infection, and the impact of the virus mutations on the infectivity. Also, they aim to predict new variants of the virus emerging in a second wave and their potential for infectivity.
With 93% identity, the COVID-19 RNA sequence is similar to a virus found in horseshoe bats (Rhinolophus Anis), but how this virus jumped to humans is unclear. Deciphering the pathways it took is key to avoid the emergence of new infections, which are in a cryptic state in other exotic animals.
The scientists aim to anticipate the virus’ next move and clarify the pathway. They want to know its mutational space, understand varying susceptibility to infection, and predict genomic changes, impacting infectiveness. The group plans molecular dynamics (MD) simulations to provide information on potential cavities in the variants of viral proteins which can be tackled by drugs.
To achieve this, the team involves four computational groups (N. López-Bigas and M. Orozco at IRB, and R. Badia and J.L. Gelpí at Barcelona Supercomputing Center) and experimental groups in Marseille and Milan. They will focus on the entrance of the virus into the human cell by binding the Spike protein to human cell receptor ACE2 and perhaps to protein CD147.
The group estimated that ± 80% of the mutations impacting infection are located in the receptor-binding region of the Spike. It is unclear how the virus will mutate when specific drugs or antibodies attack it. Or if the virus’ proofreading protein will be fully inactivated.
Through HPC study an in-depth analysis was made of mutations and potential future mutations, ranking those that are more likely to happen.
PRACE awarded the project with 6 000 000 core hours on Joliot-Curie Rome, hosted by GENCI at CEA, France.
The NANODROP project is led by Prof. Stéphane Zaleski from Sorbonne University, France. The goal of the project is to understand the mechanism of COVID’s propagation to model and prepare recommendations for protective actions.
One of the big problems of the COVID-19 pandemic is a deficit of fundamental knowledge related to generation, transport, and inhalation of pathogen-laden droplets and their pathways as airborne particles, or aerosols, in the transmission between people. In this project, scientists analyse the processes of droplet generation by exhalation, their potential transformation into airborne particles by evaporation, transport over long distances, and inhalation by the receiving host as multiphase flow processes. The team presents a simple model for the time evolution of droplet/aerosol concentration based on a theoretical analysis of the physical processes.
The group proposes a better understanding of the transmission of the virus. While gravity causes the larger droplets too quickly fall to the ground, smaller droplets will delay their fall because of viscous drag and air turbulence. They may float long distances through the air making the transmission of diseases by aerosol particles explosive.
The group cites anecdotal evidence that bursting surface bubbles in swimming pools produce film drops that propagate viruses in the neighbourhood. It is, however, difficult to estimate the number of such aerosol droplets. Moreover, sneezing is known to spread droplets at distances of more than six meters. These various effects require a better analysis of droplet formation and subsequent dispersion and evolution.
The team of Prof. Zaleski, as well as M. Herrmann’s group at Arizona, will perform new numerical investigations, and the objective is to prepare useful conclusions on the transmission of SARS by aerosols.
This framework along with new experiments and simulations of the group can be used to study a wide variety of scenarios involving breathing, talking, coughing and sneezing and in different environmental conditions, such as humid or dry atmosphere, and confined or open environments. There are still many unresolved issues around evaporation, but with a more reliable understanding of the underlying flow physics of virus transmission, an improved methodology in designing case-specific social distancing and infection control guidelines can be created.
For these calculations with the log-conform model of viscoelastic fluid in the basilisk code, PRACE awarded the project with 1 000 000 core hours on Joliot-Curie Rome, hosted by GENCI at CEA, France.
Effects of different glycosylation motifs on the structural stability and dynamics of the SARS-CoV2 S glycoprotein is led by Dr Elisa Fadda, Maynooth University, Ireland.
Discovering how to regulate activity and pathogenicity of the spike of SARS-Cov-2 is the main task of Dr Fadda’s project.
The S protein is covered with complex carbohydrates or glycans. These serve as a kind of protection for the virus as it uses them to disguise itself in order to sneak into a human cell unnoticed and deceive the immune system and host cell.
With molecular dynamics simulations, however, the game changes. Using this method, scientists can discover this invisible structure of the spike and how it reacts and rearranges when S protein binds with a human cell receptor ACE2.
Various cell hosts perform multiple types of glycosylation: the enzymatic process that attaches glycans to proteins. According to the team, this study will indicate if there are different levels of activity of the spike (CoV-2 S glycoproteins), which will be expressed in different cells.
Molecular dynamics simulations will provide insight into the functional role of the glycans in the S protein’s structure, dynamics, activation, thus discovering its strengths and weaknesses. The goal is to find out how different glycans regulate the spike’s activity.
Moreover, these crucial vulnerabilities of the spike S could be good targets for specific drugs, effective therapeutic strategies, and diagnostic interventions.
The computational resources requested from PRACE will cover the study of five CoV-2 S (spike) models, designed explicitly with different glycosylation at three specific sites. They will be consistent with available data from human cells and with recombinant protein data from collaborators (Dr R.P. De Vries from Utrecht University, The Netherlands). PRACE awarded the project with 15 840 000 core hours on Marconi100, hosted by CINECA, Italy.
COVID-RNA is a project led by Prof. Kresten Lindorff-Larsen from the University of Copenhagen, Denmark.
The team aims to predict the structure and dynamics of selected structural elements in the non-coding region of the coronavirus’ genome. Potentially, the project’s results could reveal essential relationships that are crucial in the understanding of the viral replication, transcription, and packaging and discovering weaknesses in these processes that can be exploited to combat the virus. Furthermore, the scientists will use their models as a starting point to investigate the druggability of non-coding RNA regions by small molecules. The computational work will be carried out in collaboration with partners studying these molecules in biophysical experiments.
Coronavirus genomes are single-stranded RNA of approximately 30k bases. Their genome contains protein-coding regions for the viral replication machinery, structural proteins, and accessory proteins. Additionally, untranslated regions (UTR) – 5’UTR and 3’UTR – act as elements that are decisive for viral replication, transcription, and packaging.
Despite their functional relevance, detailed information on non-coding regions is scarce.
PRACE awarded the project 20 000 000 core hours on Joliot-Curie Rome, hosted by GENCI at CEA, France and 352 000 core hours on Marconi100, hosted by CINECA, Italy.
Epi-EWS is led by Dr Alejandro Marti from Mitiga Solutions, Spain. The idea of the project is to define, adapt and refine models that help to timely detect and contain the spread of epidemics, especially for populations on the move. Of course, the model will be adjusted to COVID-19 cases. The goal is to develop an early warning system (EWS) that will be able to provide fast detection of epidemic outbreaks, accurate predictions of the disease spread, and assessment of the outbreak’s economic impact.
To achieve this general objective, it will be necessary to model the evolution of a large number of individuals related to the development (or absence thereof) of an epidemic outbreak. The scientists think that this model must take into account the habits of the individuals, their relationship with the environment, the spread of the disease, the performance of specific controls, and eventual treatment.
The experts consider such a highly complex system will be complicated to build using analytical models. For this reason, they plan to use Agent-Based Modelling and Simulation (ABMS) for simulating citizens’ behaviour. The goal is to build a comparative platform – Epi-EWS (Model Aggregator Platform for Epidemics) aimed at evaluating different ABM platforms to exercise models from the Epi-EWS project and determine which alternative is better suited to COVID-19 modelling.
Mitiga Solutions is a spin-off of the National Supercomputing Center in Spain. The company focuses on predicting and mitigating the impact of natural and social hazards. It is in collaboration with a large number of organisations around the world. The CEO of Mitiga is also a member of the Industrial Advisory Committee of PRACE.
PRACE awarded the team of Mitiga 5 000 000 core hours on Beskow, hosted by KTH-PDC in Sweden.
REDAC: REpositioned Drugs Against COVID-19 is led by Prof. Dr Vittorio Limongelli from the University of Lugano, Switzerland.
His research group is planning to exploit their decade-long experience in drug design to target some of the essential molecular players involved in the pathology of the coronavirus. Among them are the viral proteins that allow the virus to enter human cells and replicate there: Main Protease (Mpro) and RNA-dependant RNA polymerase (RdRp), and the human host proteins Angiotensin-Converting Enzyme 2 (ACE2) and Mitochondrial Assembly 1 (MAS1).
Propelled by the ambition to find fast and efficient treatment, the scientists will reposition market-approved drugs to provide easily accessible tools for the treatment of COVID-19 in multiple stages of the infection. According to them, the chosen targets cover all the phases of SARS-CoV-2 lifespan, from entry into the host cell (ACE2) to assembly of the replication machinery (Mpro), and reproduction of the viral genome (RdRp).
Drug repositioning campaigns typically lead to discovering compounds with weak activity towards the novel target. Therefore, the aim is to identify multi-target repositioned drugs towards more than one investigated goal. This knowledge will favour the rapid achievement of a clinical protocol. It will be suitable for mitigation of the infection, by slowing down the virus reproduction in the early phase and avoiding it reaching the later lethal phases.
According to the team, any data and scientific knowledge arising from the planned investigation could help in the development of novel drugs against SARS-CoV-2 and other coronaviruses.
PRACE awarded the project 350 000 node hours on Piz Daint hosted by CSCS, Switzerland.
CFDforCOVID is led by Dr Florent Duchaine from CERFACS, France. The ambition of the project is to provide an HPC tool, based on recent aerospace Computational Fluid Dynamics (CFD) solvers to simulate the dispersion, evaporation, and contagion risks of all droplets emitted by an infected person, especially in enclosed spaces such as an aircraft cabin or a car.
Also, during the project, ventilation flow, passenger locations, and the wearing of mask or not in an aircraft cabin and a car will be investigated.
The researchers estimate that droplets, due to their small size, follow a path that depends strongly on the geometry of the space in which they propagate. They can go very far from the infected individual and follow a very complex trajectory, depending on the characteristics of the airflow in the cabin or room.
The team working on the CFDforCOVID project will develop scenarios with different air conditioning specifications, with a different number of infected people breathing, speaking or coughing, and various lengths of time spent in the room. Then with CFD simulations, they will calculate the size, temperature, composition, velocity, age, and position of all virus loaded droplets at all times in three dimensions. This data will be associated with medical information (such as the lifetime of the virus in a droplet) and post-processed to provide risk evaluations. The project will assure global risk probabilities (for example, there is a 20 % chance the person sitting in this place for 45 minutes will be infected).
According to the team, such highly sophisticated approaches are a crucial element in understanding the airborne transmission of the virus. Furthermore, the findings could support science-based policies for virus control. PRACE awarded this project 10 000 000 core hours on Joliot-Curie Rome, hosted by GENCI at CEA, France.
Anti-Spike is led by Dr Miguel Soler from the Italian Institute of Technology, Italy. The main goal of the project is to exploit the available experimental knowledge to perform in silico (computer simulations) design of antibody fragments with high binding affinity to stop viral entry in the host cell. The target is the anti-ACE2 epitope of 2019-nCoV (SARS-Cov-2) RBD. ACE2 is a human membrane protein, and RBD (receptor binding site) is located on the spike of the virus that binds to ACE2. This connection is crucial for infection and scientists plan to apply the recently developed evolutionary algorithm of binder design. It has proven successful for antibody fragments and peptides as binders of protein and drug targets.
In silico maturation of the antibodies has the unique advantage that the affinity optimisation can be precisely controlled at molecular level. This technique leads to a selective binding – only for desired regions. The group thinks that antibodies, hindering the binding between the virus and ACE2, will reduce the infection and give the immune system time to react.
The researchers will optimise m396 – an antibody which is an ideal candidate for affinity maturation towards the coronavirus. In this respect, ongoing collaboration with the ICGEB of Trieste for experimental testing of the predicted antibodies will be very useful.
PRACE awarded the project 10 000 000 core hours on ARCHER, hosted by EPCC, United Kingdom.