International HPC Bank

Fueling computing liquidity with supercomputers to drive digital transformation of pandemic preparedness and response

Timothy TW Tan, Secretariat, HPC Core Group of HPC Leaders; Chief Executive, National Supercomputing Centre, Singapore; Internet Hall of Fame inaugural inductee; Medical School biochemistry professor, National University of Singapore.

Bioinformatics workflows for assessing virus variants and structural modelling of drug molecules interfering with virus infection require considerable computing power. Yet hardly any research hospital has a supercomputer. With a huge mine of patient records and digitalised medical images, all highly amenable to AI/Deep Learning, medical researchers still rely on conventional computers when the supercomputing community is achieving exascale and quantum computing supremacy. This digital divide needs to be bridged in a sustainable way. 

Currently, the international research and health sectors cannot achieve this as a community due to the feast-famine nature of the infrastructure upgrade-cycles. Amongst developed economies with national supercomputing capabilities, on average, half have no flexibility to respond to crisis-related demands, being in the latter part of their lifecycle where their machine is over-committed with no spare capacity. The other half have excess capacity, being in the earlier part of their lifecycle, where the new machine has more capacity than the local research sector can effectively utilize. These unutilized CPU-hours can be deployed for pandemic response workloads as healthcare communities increasingly exploit AI/Deep Learning and Big Data analytics; but only if mechanisms for capacity sharing are hammered out by the national supercomputing facilities and their government funders.

Our solution serves humanity in the broadest sense by facilitating more efficient, effective pandemic preparedness and response within an internationally democratized HPC framework. Few countries have systematic biosurveillance to detect changes in the animal reservoirs for zoonotic leaps. While sequencing costs plunge, we are still not doing metagenomic baseline environmental monitoring and developing early warning systems. Our sequencing pipelines are not connected to data lakes and our medical health systems. Where systems  have good data and information flow, we have not yet coupled them to diagnostics development pipelines that can scale logistically to millions of kits. In drug screening, for developing new drugs for treating new diseases like Covid-19 and repurposing established drugs through rapid screening using in silico techniques, the digital infrastructure is still found wanting. In vaccine development, despite the unprecedented speed of emergency use vaccine, mutant strain monitoring and phylogenetics modelling from the bioinformatics community is not coupled tightly enough to the diagnostic pipeline, and not yet fast enough to feed new sequences into our biomanufacturing pipelines of mRNA vaccines against variants. We can do better because the technology is available computationally and the computing and networking infrastructure ready for this level of tight integration. 

The singular public good would be that our HPC Bank effort will evolve into a prototypic computational framework for the support of BioDefence organisations worldwide. It is like the Internet 25 years ago, where by now, an academic research network has now emerged into a global information infrastructure.  We could not have survived this pandemic without the internet not melting down while delivering all our Zoom meetings and our Netflix movies to maintain our work efficiencies and our sanity during lockdowns. So also, we envisage that HPC Bank will provide the framework to build our biodefence information system to facilitate a command, control, communications, computational and computing speed in the face of any new pandemics, including the mitigation of our current one.

Our HPC Bank will help our own supercomputing communities in flattening the feast-and-famine cycles, but there will be sufficient spare resources for us to expand the scope of assistance to our life science communities in particular to facilitate their impact and transformational change. By bridging this digital divide it will help bolster our case and expand our footprint.

Our collective user base already exceeds tens of thousands, and in turn, their research work impacts millions in four continents for starters. Starting with the NCI Australia / NSCC Singapore beta trial and expanding in partnership with the Trinity Challenge across our consortium of tier1 national and tier0 multinational facilities, we can integrate and target them for pandemic preparedness now; for Disease X, the impact of such an effort would herald a new era of biodefence forces worldwide putting together resources for a common defence system against any future emerging disease of pandemic potential.

Measurable indicators would include:
a. number of supercomputing centres willing to share our vision of pooling resources in this novel way

b. number of biomedical institutions and medical research hospitals which we start negotiating with to implement HPC solutions in AI and Big Data Analytics and High Throughput Computing for their health data and anonymised health records

c. number of conferences/mtgs/workshops organised and talks we deliver on the HPC bank

d. amount of publicity and stories describing our efforts

e. amount CPU-hours and GPU-hours, PetaBytes of storage committed and network connectivity to other supercomputing centres participating

Barriers include:

a. Resistance by HPC policy makers in sharing resources paid for by tax-payers of their respective jurisdictions

b. Issues of data sovereignty

c. Cybersecurity and data confidentiality which hospital systems are particularly sensitive to

We believe we have sufficient influence to mitigate some of these issues and overcome these barriers to a sufficient level to trigger a wave of change.

National Supercomputing Centre Singapore (TTWTan); National Computational Infrastructure at the Australian National University (Sean Smith); CSC-IT Finland (Kimmo Koski); RIKEN, Japan (Satoshi Matsuoka); University of Warsaw Poland (Marek Michalewicz); NECTEC Thailand (Piyawut Srichaikul); Pawsey Supercomputing Centre, CSIRO Australia (Mark Stickells).

We need the validation of the HPC Bank concept and what better way than to apply the concept to stimulating solutions to the pandemic preparedness problem with solid supercomputing infrastructural and resource support. To receive endorsement from The Trinity Challenge will help to raise the profile and credibility of our efforts, and encourage other supercomputing centre leaders to join in this effort to establish a more lasting solution to what has already started with Covid-19. Their donation or contribution or deposit of HPC resources will help supercomputing centres in need, and biomedical organisations within their sphere of support to draw upon. Similarly funding from The Trinity Challenge will certainly help defray the costs of resources dedicated to helping winning projects of the Challenge.

We wish to partner The Trinity Challenge participants and to support each of them since they have a strong digital agenda and scaling of their solutions to support their services with supercomputing power will certainly help them increase their chances of success.