NCI Australia and the Pawsey Supercomputing Centre are supporting the Australian and international research community undertaking COVID-19 research through provision of streamlined, prioritised and expedited access to computation and data resources.
Both national facilities are contributing resources to support researchers in Australia in the fight against COVID-19. NCI today announces support for three targeted projects with over 40 million units of compute time, which is equivalent to one single computer doing constant calculations for over 4,500 years, on the Gadi supercomputer; while Pawsey Supercomputer Centre has provided access for researchers across five projects to over 1100 cores on the new deployed Nimbus cloud.
NCI’s Director Professor Sean Smith said that the successful grant recipients will have access to compute power that’s never before been available in Australia.
“The Australasian Leadership Computing Grants represent a peak merit award for high-performance computing at a scale that is unprecedented within the Australian context, made possible by the Federal Government NCRIS funding for the new peak facility recently commissioned at NCI,” Professor Smith said.
“We look forward to working with Australian researchers to deliver this prestigious grant and trust that their projects will ultimately contribute substantially to the global scientific effort addressing the COVID-19 pandemic.”
In addition to access to the Gadi supercomputer, NCI will provide storage for major reference data sets (national or international) needed to support the computational and data analysis by multiple projects.
Resources being made available at the Pawsey Supercomputing Centre include large data stores and the new refreshed Nimbus Cloud service. COVID-19 projects will have access to allocations of up to 500 cores, with up to 100 TB of storage available for use during 2020 – 2021.
The rapid deployability of the Cloud resources means that Pawsey staff have been working with researchers from The University of Western Australia, Monash University, WA Department of Health, the Queensland Facility for Advanced Bioinformatics and Telethon Kids Institute for the last two weeks.
“In response to COVID 19, rapidly advancing the relevant science can deliver real public health outcomes and this initiative is all about rapid and expert support at a time of great need” said Mark Stickells, Pawsey Executive Director.
“Having access to advanced HPC resources and data expertise at Pawsey and NCI allows Australian researchers to accelerate their science to combat the pandemic and we are proud to contribute our national infrastructure and expertise in this collaborative effort.”
Research projects gained access mostly to Pawsey Nimbus allocations, however after discussions with researchers other needs where identify that included training, and GPU resources on Topaz cluster, a new commodity Linux cluster, supporting pre- and post-processing of data, throughput workflows, and simulations.
NCI Australia and the Pawsey Supercomputing Centre are Australia’s leading supercomputing facilities. NCI Australia is based at ANU in Canberra and supported by ANU, the Bureau of Meteorology, CSIRO, Geoscience Australia and the ARC. Pawsey, in Perth, is a joint venture between CSIRO and Western Australia’s four universities.
The following three ALCG-COVID-19 awarded projects will begin computation on NCI’s Gadi by 30 April 2020.
Further information can be found at www.nci.org.au.
Associate Professor Megan O’Mara
Dr Katie Wilson
Dr Stephen Fairweather
The Australian National University
Using large-scale molecular dynamics for rational drug design
This research uses simulations of the around 800,000 atoms that make up a key receptor of the human body to understand exactly how the coronavirus uses it to invade human cells. It is only with high-resolution modelling that accurately replicates the true behaviours of these receptors that we can figure out where vulnerabilities in the virus’ binding process are. Targeting the interaction between the human receptors and the coronavirus binding protein might well be a useful direction for drug design. This project will produce world-first vital information about regions of the receptors that could be potential vaccine or drug targets. Using 48 processors running for 19 days for each of 64 molecular simulations, this research will spend around 13 million units of computing time in the coming months. This amount of high-performance computing is only available on NCI’s new Gadi supercomputer.
Associate Professor Michael Inouye
Dr Sergio Ruiz Carmona
Baker Heart and Diabetes Institute
Structure-based drug discovery
This project aims to model existing or novel drugs, along with drug binding sites, that could be used in treating COVID-19. As a first step, more than 7 million existing drug compounds from a global repository will be modelled to find ones that may inhibit a key step in the virus replication process. The researchers will focus in on likely candidates through a sequence of molecular dynamics simulations. Expanding further to include more proteins involved in the virus binding process, the researchers will also analyse the complex interactions between the virus and human proteins. Identifying key interactions involved in the virus binding process will help to define the kind of drug that could be used in treating it. Over the rest of 2020 and into 2021, the researchers plan to spend around 12 million hours of computing time conducting their simulations.
Professor Alan E. Mark
Dr Martin Stroet
Ms Shelley Barfoot
University of Queensland
Targeting structural transitions in the COVID fusion protein
By modelling key structural changes in the protein that enables a coronavirus to enter and infect a human cell, this research project aims to throw light on a critical target for both vaccine development and the discovery of antiviral agents. In particular, working with researchers from one of three centres worldwide charged with rapid vaccine development by the World Health Organisation, the aim is to understand how potential vaccine constructs that incorporate the coronavirus fusion protein can be optimised. Professor Mark and his team will also use the national supercomputer facilities to assist in the global effort to identify existing drugs that could be repurposed to treat COVID-19. The Automated Topology Builder (atb.uq.edu.au), a globally recognised molecular modelling tool developed at The University of Queensland, will be used to develop high quality atomic interaction parameters for all pharmaceutically active compounds that have passed phase 2 clinical trials (efficacy and safety). These will be made freely accessible to researchers at universities and public research institutes worldwide. High quality atomic interaction parameters are needed by researchers trying to identify which currently available drugs might be effective in treating COVID-19 disease. Expanding the number of molecules for which high quality parameters are publicly available will also have long-term benefits in structural biology, computational drug design, materials science and more.
The following projects are already benefiting from Pawsey allocations:
Associate Professor Michael Wise
The University of Western Australia
Are SARS and SARS-Cov-2 Subclasses of Bat Coronavirus?
The team has been allocated resources on the Nimbus cloud to work on a phylogenetic analysis of 75 genomes from SARS-CoV-2, SARS and bat(+civet) CoV – to understand the parentage/lineage of those genomes and their similarities.
Dr Tom Karagiannis
Molecular modelling COVID-19 targets
The team has been allocated GPU resources on Topaz for COVID-19 targets molecular modelling. In the absence of a vaccine, one of the key immediate research areas is repurposing existing compounds with potential antiviral effects, the research project aims to provide a molecular basis for known antivirals and identify any new ligands which may offer a protective effect.
Telethon Kids Institute
Bayesian Adaptive Trials for COVID-19 Response
The team has been allocated resources on the Nimbus cloud to develop Bayesian adaptive clinical trials (BATs) for COVID-19 research projects helping to make statistical inference about potential treatment outcomes. Each trial involves simulating 1000s of participants, randomisation, interim and final analyses, repeated 1000s of times for 100s of scenarios.
Dr Gareth Price
Queensland Facility for Advanced Bioinformatics – University of Queensland
Melbourne Bioinformatics – University of Melbourne
Galaxy Australia COVID-19 dedicated Pulsar
Galaxy Australia relies on remote (to head node) deployments called Pulsar to increase the range and number of jobs that can be run on the service. The team has been allocated resources on the Nimbus cloud to deploy a dedicated COVID-19 Pulsar as part of Galaxy Australia at the Pawsey Centre that allows Galaxy users to rapidly analyse their data on published tools/workflows to further research into SARS-CoV-2.
Researchers can still apply for resources at Pawsey, since this special call is ongoing.
Researchers requiring Pawsey resources should visit the Cloud application portal, marking their application with “COVID-19”: https://apply.pawsey.org.au/
Individuals with no Australian Access Federation (AAF) credentials required to log in on the Cloud portal are not excluded and should contact email@example.com for further support.