Supercomputers are key tools in searching for responses, defence, treatments and antidotes for COVID-19. In this post I look at the publicised pandemic research activities associated with these high performance systems.
What research is needed?
We can break the current scientific research being done into specific areas. For instance:
- Genetic analysis of the COVID-19 virus spike protein
- Detailed analysis of compounds used to counter the affects of the virus on patients
- The creation of effective tests to detect those suffering from the virus and the antibodies of those that have recovered from it
- The creation and digital testing of anti-viruses
- Demographic analysis of the spread of infection and contact tracing
- Modelling the socio-economic affects of the pandemic
Specific supercomputer activities include molecular dynamics used to define the 3D structures of the protein target, which in he case of COVID-19 has a Spike structure, for which scientist have developed the GROMACS code. Another application is molecular docking, which attempts to bind drug candidate molecules to the target protein digitally.
How national and regional supercomputers are being harnessed
America
The US federal government has set up the Covid-19 HPC Consortium including the Energy Department’s 17 national laboratories, including Oak Ridge (home to IBM’s Summit supercomputer), Lawrence Livermore (home to Sierra), Los Alamos, Sandia and Argonne National Laboratories. In total the 16 supercomputers housed in these facilities have a capacity of 330 petaFLOPs in total.
Their objective is to investigate the epidemic, characteristics of the virus and to create potential vaccines and drug treatments. They have already examined 8,000 compounds, narrowing them down to 77 thought to be most useful. In the Argonne Lab 250m known small molecules that affect the virus have been examined.
Computer companies involved in helping out include IBM, Microsoft and Google. Universities involved include MIT and the Rensselaer Polytechnic Institute. Researchers are submitting proposals through a central portal, allowing the steering committee to connect researchers to the appropriate supercomputer resources.
Asia Pacific
In China supercomputers are being used to help develop potential vaccines. It is also offering doctors across the globe free access to an AI tool for early diagnosis of those affected with the disease through analysing chest x-rays. Beyond supercomputing tech companies have been providing autonomous vehicles to take supplies to medical workers, fitted drones with thermal cameras to detect those with the disease and others to spray disinfectant.
In Taiwan the Taiwan Computing Cloud (TWCC) includes the 9 petaFLOPs Tawania 2 supercomputer. It is being set aside for projects including medical research, disease prevention, revitalising the economy, helping online education and other activities. It has decided not to take on projects originating from mainland China however.
In Japan the RIKEN-led supercomputer Fuhaku is being installed in Kobe to counter the epidemic. Although the server won’t be fully up and running until 2021 some trial nodes are already in use. Research projects into understanding the characteristics of the virus, identifying compounds that can be use as therapeutic agents, improve diagnosis and treatment, give insights into the socio-economic impact and others into countermeasures against the disease.
Researchers in the Peter Doherty Institute for Infection and Immunity in Australia have studied how the immune system reacts to the virus by analysing the blood of patients during the infection, which will help in choosing candidate treatments.
Europe
In Italy Common Infrastructure for National Cohorts in Europe, Canada, and Africa (CINECA) houses the 10 petaflops Marconi supercomputer, which was used as part of the EU‘s ANTAREX project to investigate the Zika virus. It has also been selected as one of the sites for the Leonardo supercomputer to be installed in 2021 for full production in 2022. It is now one resource for the EU’s Exscalate project funded to help researchers testing scientific solutions to COVID-19. Other sites involved in the project include the Barcelona Supercomputing Project in Spain, Julich Supercomputing Center (Germany) and KTH supercomputing centre (Sweden).
The UK government is funding 6 separate projects to investigate potential treatments to those suffering from the disease including one at the University of Oxford of a potential vaccine made from a harmless adenovirus altered to produce the spike protein of the coronavirus after injection.
Vendor approaches
IBM, Alibaba and quantum computer specialist D-Wave have offered their computers and AI technology to researchers. The Vodafone Foundation has partnered with Imperial College in the UK to enable users who download the DreamLab app to lend their smart-phones to act as a virtual supercomputer for researchers investigating food and drug molecular discovery in phase one before going on to create treatments and provide nutritional advice.
Would a more co-ordinated approach be more effective?
Just as individual government reactions to social distancing, lock-down to protect citizens and delay the rapid spread of infection to prevent healthcare institutions from being overcome differ, it seems to me that currently supercomputers are being used in a blunderbuss way with many countries trying to address all applications at the same time. Although the competition between national researchers may spur on some to major breakthroughs, there is a lot of duplicated effort; a more targeted approach in which individual countries addressed a narrower set of objectives might produce faster results.
However it’s reassuring to know that some many scientists are working fervently on essential medical research and that others are looking at how we can recover economically when things return to normal. The more sharing of information and shelving of national political issues the better.