Skip to main content

Using supercomputing for SARS-CoV-2 genome sequencing

Supercomputing Wales (SCW) has taken part in crucial work that has been undertaken at Cardiff University to support the UK pandemic response. This has included supporting the COVID-19 Genomics (COG-UK) consortium, which received in excess of £30M of government funding to establish SARS-CoV-2 genome sequencing in the UK (with a grant of over £1M for Cardiff University).  

The University hosts one of the two computational sites that provide platforms for data aggregation and analysis of SARS-CoV-2 data in real time. The SCW team has supported the setup of these systems, contributing expertise in a range of areas – from completing work to provide assurance around the security of the system through to supporting the on-lining of new capacity to enable the processing and storage of more than 500,000 SARS-CoV-2 genomes in the last 15 months. This support is now transitioning to providing analysis capability directly to the UK Public Health agencies, as well as continuing to support the research mission of COG-UK. The University’s contribution also includes bioinformatics and sequencing work undertaken by members of the School of Biosciences.  

This work has impacts at multiple levels. At a local level, work undertaken in partnership with Public Health Wales (PHW) has contributed to multiple outbreak investigations in Wales, at a regional level, the data being generated underpins the COVID-19 genomic surveillance efforts in Wales. At a national level, the team at Cardiff and PHW has generated analyses for the Welsh Government’s Technical Advisory Group and has directly contributed to multiple COG-UK reports presented to UK Government SAGE.   

“I’ve been working on using patient metadata datasets to investigate the geographic spread of lineages of SARS-CoV-2 within Wales, and the impact of the importation of lineages into Wales,” says Dr. Anna Price (Research Software Engineer, Supercomputing Wales and CLIMB-BIG-DATA). “This work feeds into the COVID-19 genomic surveillance efforts in Wales and would not be possible without the infrastructure at Cardiff and the datasets that COG-UK provides, as well as the skillset I bring as a RSE in being able to generate analyses of these datasets.”  

Supercomputing Wales has provided essential support to this work through hardware provision and the expertise of the SCW team. Key contributors at Cardiff University include: Prof. Tom Connor (Professor, Cardiff University; Bioinformatics Lead, Public Health Wales), Dr. Anna Price (Research Software Engineer, Supercomputing Wales and CLIMB-BIG-DATA), Alex Southgate (Systems Engineer, CLIMB-BIG-DATA), Drew Mack (PhD student, CDTAIMLAC), Dr Christine Kitchen (Supercomputing Wales), Professor Martyn Guest (Supercomputing Wales), Steffan Adams (Supercomputing Wales) and Robert Munn (Supercomputing Wales). The group also recognises the support of the datacentre team who have facilitated datacentre works required to support the pandemic response activities.  

Global Mangrove Watch – Supercomputing for keeping a watchful eye on Mangrove Forests

Mangroves – forests that sit where the ocean meets land – are key to supporting the natural world and effective climate action. Researchers at Aberystwyth University are part of the major project Global Mangrove Watch, initially founded by Prof Richard Lucas (Aberystwyth University) and Ake Rosenqvist (soloEO) and recently expanded with the Global Mangrove Alliance, which seeks to bring diverse stakeholders together towards a common goal of conserving and restoring mangrove ecosystems. 

Dr Pete Bunting is a Reader in Remote Sensing within the Department of Geography and Earth Sciences. He’s put his expertise in satellite earth observation at the service of Global Mangrove Watch (GMW) and, by using Supercomputing Wales resources to store and process huge datasets, has been able to ensure that GMW can offer the best possible maps of the planet’s mangroves [1]. 

“Mangroves are a vital part of the earth’s ecosystem on many levels,” Dr Bunting says. “For example, they grow where other vegetation cannot, protecting coastlines and human habitations from storm surges, and capturing and storing carbon, which helps mitigate rising atmospheric CO2 levels.” 

“The GMW online portal not only maps the extent of mangroves on an annual basis but also offers users – which might be governments, local authorities and environmental groups – near real-time alerts to changes in mangrove conditions. This allows users to become aware of changes at an earlier stage, so they can take action sooner.” 

Dr Bunting’s contribution to the GMW initiative has some high-profile users, with the data being the official mangrove data selected by the United Nations’ Environment Programme (UNEP) for reporting on Sustainable Development Goal, Indicator 6.6.1 [2], as well as in use on the UN Ocean+ Habitats platform [3], which aims to provide the world’s decision-makers and communities of practice with the best possible information and tools for ocean ecosystem management and conservation. It is also the mangrove layer used by the World Resources Institute on their Global Forest Watch and Resource Watch platforms [4, 5]. The GMW is being used to inform the European Space Agency’s Climate Change Initiative Biomass Project [6]. The health of ocean ecosystems has important social and economic as well as environmental effects and Dr Bunting’s work in this area has been submitted as an Impact Case Study in the 2021 Research Excellence Framework (REF), the system for assessing the quality of research in UK higher education institutions. 

Participating in far-reaching and impactful global schemes puts Wales-based research on the map, and means that quality is key. The high-resolution satellite images that Dr Bunting and his team collect cover large geographic areas and are therefore very large data files. The SCW high-performance computing facility allows this large-scale data to be processed quickly and efficiently. “Mangroves are constantly evolving and policy and decision-makers need to be able to access high quality information quickly and easily. This is particularly valuable in the developing world, where many countries don’t have the resources to be able to develop their own mangrove maps and information systems,” says Dr Bunting. “We’ve processed and classified about half a million images on the supercomputer so far – doing this would take years and years on a desktop or laptop computer. In short: without HPC, our work would be impossible.”  

GMW datasets are available for download at 

Analysing 20 years of data on the atmosphere of the sun

The solar system physics team at Aberystwyth University uses Supercomputing Wales to study the sun’s atmosphere, analysing data collected over the past two decades by a range of satellites and ground-based telescopes.

The sun’s atmosphere is still “quite mysterious, with a lot we don’t understand,” says Dr Huw Morgan.

“What Supercomputing Wales allows us to do is to analyse long periods of data, up to 20 years’ worth of data, in quite a short period. We have around 15 terabytes of data stored on Supercomputing Wales and that data can be analysed in several ways, so there are different processing streams depending on what we are trying to study,” Morgan says.

One current field of study is the shape of the solar atmosphere.

“We have two dimensional images, but we don’t know the three-dimensional shape. So one of the most advanced methods we’ve developed is to use tomography, similar to medical tomography in that you take images from many different angles and work out the three dimensional shape of what you’re looking at. That involves taking thousands of images over the course of a month, as the sun rotates,” Morgan says.

The work Morgan’s team runs on Supercomputing Wales differs from the heavily parallelised code run by many other physics departments, he says.

“Someone doing quantum physics really needs the parallel processing power of Supercomputing Wales. We do run jobs in parallel, but running one year of data in one job, another in a second and so on – so we can process 20 years of data in one 20th of the time. It’s totally invaluable to our work – we wouldn’t be doing the research we’re doing without Supercomputing Wales,” says Morgan.

The solar system physics team works closely with Supercomputing Wales support staff. Colin Sauze, a Research Software Engineer for the facility is based in the office upstairs from Morgan and regularly helps the team out with problems, or runs workshops on specific issues.

Access to the facility has never been an issue, with computing time usually available within “seconds, or minutes,” Morgan says.

Modelling the power of the sea as a low-carbon energy source

Using the power of the sea is nothing new: people have been using tidal range – the rise and fall of sea levels – since medieval times to drive mills and grind grain. New technologies now allow us convert tidal and wave power into electricity, and research from the School of Ocean Sciences at Bangor University aims to understand the available resources and how they can best be harnessed.

“The good thing about tides is they’re predictable, so you can see what contribution they can make for a hundred years into the future. There’s the tidal range, and there’s tidal stream, where you harness the kinetic energy of the tide using a horizontal axis turbine, like an underwater windmill,” says Simon Neill, Professor in Physical Oceanography.

“And then we have wave energy. It’s less advanced than tidal energy because waves by their very nature are quite destructive, and of course you want to put your devices somewhere where the waves are energetic – so at present the devices have to be really overengineered to withstand those forces,” says Neill.

Neill’s research focuses on understanding the resource itself, and identifying the optimal locations for energy devices.

“Waves, for example vary over a long timescale, and you can’t really say anything meaningful about the resource until you have at least a ten-year record. So you could go and put a wave measuring device in place and wait ten years… or you can use modelling. We use measurements to build a model and see how things will vary over 10, 20 or even 50 years. You can look at things like climate change: no one really knows what the wave climate will look like in 2050, so the only thing you can do is run a model,” he says.

The team also looks at the interaction between energy extraction and the resource.

“If I put a 100 megawatt array in place, I wouldn’t really know what the impact on the environment would be until post construction. By modelling you can see what the impacts would be, and change the shape or the spacing between devices to minimise impacts on the environment,” Neill says.

Tidal, and especially wave models, are “extremely computationally expensive” and can take up to three months to complete, even on a facility as extensive as Supercomputing Wales, but the facility and team make it easy for Neill.

“Before this we had to spend a lot of our time setting up and optimising our models – and that was just taking away from the science. Having this dedicated team who deal with all the technical issues is great. After some discussions about a new model application, they’ll take care of all the technical set up and optimisation, and provide me with some scripts to run the model, so I can focus getting outputs,” he says.

Studying the impact of the next generation of nuclear power system materials

The environment inside nuclear systems is, unsurprisingly, extremely harsh. Even materials that work well in other difficult environments, like aerospace applications, can degrade very quickly in a nuclear reactor.

Simon Middleburgh, Reader in Nuclear Materials in the Nuclear Futures Institute at Bangor University is using the Supercomputing Wales cluster to study the impact of the next generation of materials for nuclear power systems and also gaining a better mechanistic understanding of the current generation of materials to improve their reliability.

“Instead of making expensive mistakes by putting novel but un-tested materials inside nuclear reactors, we do simulations on them beforehand so that we understand how they are likely to behave,” Middleburgh says.

Middleburgh’s group is involved in an international initiative devoted to developing Accident Tolerant Fuels, aiming to develop new fuels that are both safer and more economically viable, and is also working on next generation energy plants using alternative coolants.

“We use Supercomputing Wales to help us make the first steps in designing the materials, so that we can understand things that are happening at the atomic scale, things like radiation damage effects.

“When you hit a material with a neutron it’s like a break-off shot in snooker – in the ideal material, you hit the snooker balls (atoms) and after knocking into each other and transferring all of their kinetic energy they end up in the same triangle back in the middle of the table! Lots of materials do this, but you have to be careful not to choose the ones that don’t,” he says.

Computer simulation is vital in gaining a mechanistic overview of what is happening within the materials because any mistakes are so serious, and predicting unknowns is important for improving safety in what Middleburgh stresses is already an incredibly safety conscious industry.

Research Software Engineers at Supercomputing Wales help the Bangor project team to keep their code up to date and working optimally.

“Whenever there are new nodes and things like that, they do the testing and make sure it all works. From our end, all we need to worry about is the input files – they tell us the best way to submit them. We’ve never had any problems, it’s a super team,” Middleburgh says.

“Like a lot of things in academia, you can have the best mind in the world, and the best ideas in the world, but if you don’t have the equipment that’s what stops you making progress. And that’s not the case here. It levels the playing field in Wales, which is fantastic.”

Measuring prejudice in autonomous agents

Roger Whitaker, Professor of Collective Intelligence at Cardiff University and Academic Director of Supercomputing Wales, has been working in collaboration with MIT on research which suggests that prejudice can easily develop in populations of autonomous agents. This has implications for understanding prejudice, and the development of

prejudicial groups, in humans.

“We were trying to understand what makes simple agents or bots become prejudicial, and whether it is a phenomenon that exists in nature – are there pressures on a population to become prejudicial. So it’s quite abstract, and fundamental, but it requires a lot of supercomputing power,” Whitaker says.

Whitaker’s team set up a model to test a cooperation problem, where agents have to interact and decide if they’re going to help each other.

“It’s called indirect reciprocity and it’s seen a lot in everyday life. Holding a door open for someone, for example: there’s a small cost to us, but a greater benefit for the other person,” he says.

“We set this up as an artificial life problem where agents have to decide whether to donate to other agents. We set it up to examine whether agents become prejudicial: whether they would start to decide not to cooperate based on the recipient not being in the same group as the donor,” he says.

Using Supercomputing Wales, the team was able to run “thousands of agents playing together in simulated donation games where the agents had the opportunity to form groups based on prejudice”, he says.

“We were able to control the experimental parameters, and observe conditions where agents with similar prejudicial views formed groups, through what is called homophily. Agents with a similar prejudicial disposition were assumed to have a higher chance of assimilating with one another based on being comfortable with each other’s behaviour.

“We found that there are forces that seem to make it easy for individuals to become prejudicial and to come together in groups. Within those groups, cooperation emerges – so they become highly cooperating but isolated units,”

Whitaker says.

“It also showed that when you get systems of devices that have some degree of autonomy and sensing, there is the potential for prejudice to take hold of its own accord,” he says.

Working with MIT helped to raise the profile of the project, Whitaker says, and was a good chance to show what Supercomputing Wales offers.

“This international collaboration showed that Wales has cutting edge facilities and capabilities, including the capacity to accelerate research through highly skilled research software engineers. This environment allowed the work to reach its full potential” he says.

“These models are heavily dependent on random choices by agents. They’re also evolutionary: over generations, agents need to start interacting and then periodically stop and decide whether to change behaviour based on social learning. The supercomputer allows us to play out scenarios and run the same simulations again and again, with different random starting points. Our RSEs are able to help optimise that, and have a background to contribute to the simulations, understanding the concepts and suggesting how to translate them for maximum computational efficiency.”

Sequencing the DNA of cancer and genetic diseases

Wales Gene Park, embedded within the School of Medicine at Cardiff University, provides DNA sequencing facilities for cancer and rare genetic disease researchers.

“Cancer research colleagues may be interested, for example, in sequencing the DNA of a tumour and comparing it with the individual’s blood DNA to identify aberrant mutations, or rare genetic disease researchers might be interested in particular mutations within particular genes,” says Data Strategy and IT Infrastructure Lead Kevin Ashelford.

“We also look at the expression of those genes. We each have a whole portfolio of genes that are switched on and off to be expressed at different levels within each of our cells. By profiling that expression, we can assist colleagues in exploring different diseases,” he says.

This research raises challenges due to the amounts of data involved. When an individual’s DNA is sequenced it produces vast files full of ‘sequence reads’, or DNA fragments, which need to be interpreted by comparing them to the reference genome – the human genome which was mapped in 2003.

“That requires sufficient computer processing power and memory to take all of those reads and map them, to produce a resulting alignment that we can then examine further. Once we have done that, we need to interpret the genome that we have put together, and then we need to visualise it. So there are various stages where the compute power and storage capability of Supercomputing Wales are required,” Ashelford says.

“At Wales Gene Park we do have staff who are familiar with this technology and so we worked with Supercomputing Wales to develop our own separate partition on the system, dedicated to our requirements. It’s a collaboration, really. They provide the essential IT engineering system administration skillsets and we bring the data science skills needed to run the software,” says Ashelford.

Supercomputing Wales RSE Research Software Engineer Anna Price has also worked with Wales Gene Park on a natural language processing algorithm that has simplified the task of curating information from research papers.

“A lot of research papers are being generated with details of medically important mutations, and the Human Gene Mutation Database, based here in Cardiff, gathers that information together for clinicians and researchers, providing the largest resource for finding disease-causing mutations in the world. They’ve always done it manually, but there’s a huge amount of papers to go through. So Anna took some natural language processing algorithms and applied them to the problem. Her program identifies those papers that are likely to be of interest, and flags them – so it’s not eliminating the need for a curator, but certainly begins to automate the process.”

Modelling the impact of Covid-19 on Wales

Mike Gravenor is a professor of epidemiology within the medical school at Swansea University, and a member of the Welsh Government’s Technical Advisory Cell.

When the Welsh Government realised it needed locally-relevant modelling of Covid-19 developments, it contacted Gravenor to ask what was possible.

“Modelling has played a very prominent role in trying to understand the epidemic and to plan control policies. And these models are very, very detailed, including information for the whole of the UK. So we had access to that output but we wanted to run slightly different scenarios for Wales, and use some of the additional information we had for Wales.

“I’ve got a background in epidemiological modelling, but not models of this scale, and the timescale they were asking was pretty frightening! So I contacted Biagio Lucini, Swansea University’s PI for Supercomputing Wales, and asked for help,” Gravenor says.

Lucini opened a new project to investigate the possibility of running existing modelling software on the Supercomputing Wales cluster, and adapting them for Wales-specific scenarios.

Using Supercomputing Wales has been essential to the process, Gravenor says.

“We still have a lot of uncertainty about this disease and how it spreads. We understand the general processes, but to try to answer precise policy questions, such as what happens if you allow certain activities or open schools in certain way, the outcome depends on a huge amount of unknowns. If every time you run the model you have to run it for many, many thousands of parameter combinations, that can only be rationalised by some pretty powerful computing.”

Senior Research Software Engineer Mark Dawson took on much of the work in adapting the code to Supercomputing Wales.

“Historically, someone in Mike’s position would just have been given access to a supercomputer and the rest would be down to them. But researchers are experts in their specific domain: they can’t always be expected to also be experts in supercomputing. Having the support and expertise of Research Software Engineers is about providing computational expertise to researchers, so they’ve got the tools they need to explore new and exciting territory.” he says.

Dawson has developed the code to a level where researchers can run scenarios themselves and quickly answer questions from the Welsh government.

Feedback from Government has been positive, Gravenor says, due to the fast turnaround from Supercomputing Wales.

“To get the models running within days, then build in more and more Wales-specific data – they’ve been really happy with the progress we’ve shown.”

Update May 2021:

Dr Ben Thorpe, Dr Mark Dawson, and Dr Ed Bennett of the Supercomputing Wales Research Software Engineer team have continued to work closely with Prof. Biagio Lucini, Professor of Mathematics and Supercomputing Wales Swansea Principal Investigator, and Prof Mike Gravenor, Professor of Epidemiology and Biostatistics at the Swansea University Medical School, to support the Welsh Government’s efforts to model and understand the spread of COVID-19 in Wales. This has included developing and optimising an epidemiological model specific to the Welsh context, and applying that model to produce estimates of how the spread may change based on changes in policy and behaviour. Results of these predictions formed a significant part of the Reasonable Worst Case analysis published by the Welsh Government’s Technical Advisory Cell, and drove the decision making around the firebreak lockdown in October–November 2020 and the early move to tier four restrictions before Christmas 2020. This work takes advantage of the RSE team’s unique skillset to rapidly develop, modify, and deploy software to respond to the changing real-world situation, and also relies on the Supercomputing Wales High-Performance Computing infrastructure in Swansea to run the analysis at speed, allowing it to complete in time to inform ministerial decisions. The hardware resources and Research Software Engineering team provided by Supercomputing Wales were a key enabler for this activity; had these resources not been available, then this vital work would not have been possible. 


Calculating the properties of catalysts for chemical development

Current research by the School of Chemistry at Cardiff University involves looking at ways to convert methane gas to methanol to make it easier to transport, at how to covert glycerol – a side product of biodiesel production – into new, usable products, and at whether it is possible to convert CO2 into methanol and so contribute to a reduction in global warming by using carbon dioxide in manufacturing.

“We try to work out at atomistic level how reactions take place on the surfaces of catalysts, and why you need certain metals or oxides to do the catalysis for you. We look at how the molecules absorb using the computer, and then compare that to experiments in the academic labs,” says Dr David Willock.

The team uses Supercomputing Wales as well as other UK facilities including ARCHER in Edinburgh, Isambard in Bristol and Thomas at UCL in London.

“We require a lot of calculation power. If we’re trying to look at a few hundred atoms we run generally about 200 cores at a time, so we need access to a supercomputer that can give us a slot when we need it,” Willock says.

To use ARCHER and Thomas, Willock has to submit a bid every six months and is granted a set amount of computer power on each.

“Supercomputing Wales, on the other hand, is free at the point of use for academics in Wales, so we can just try things on Supercomputing Wales that we can’t afford to try on the other facilities. We can use it as a test bed and then, if we’ve got a really big calculation to do, we’ll take it to ARCHER,” he says.

Work is completed faster on ARCHER, but the accessibility of Supercomputing Wales makes it invaluable, he says.

“And Supercomputing Wales isn’t just the machine – it’s the people that help you run on it. We don’t compile the code we use by ourselves, for example; we give it to the Supercomputing Wales RSEs to compile for us. They get the code up and running efficiently, they know all the best libraries to use and compiler options for their hardware. It would take us a long time to set it up ourselves.

“They also help us with any particular scripts for defining the chemical problem on the supercomputer that we’ve written ourselves, which sometimes need a bit of tweaking. We have a good relationship and the team has always been really responsive,” Willock says.

Harnessing the power of supercomputing to map plant genomes

The genome of a plant is just as complicated as that of a human, and its analysis requires enormous amounts of computing power, says Dr Yuan Fu, post-doctoral research assistant in plant bioinformatics at Aberystwyth University.

Fu is part of the Biotechnology and Biological Science (BBSRC)-funded Cores Strategic Programme (CSP) in Grasslands and Crops for Challenging Environments project team, studying the genome of important forage, amenity and bio-energy grasses, including perennial ryegrass (Lolium perenne). Through funding from the Global Challenges Research Fund, Fu and colleagues have been applying their expertise in ryegrass genomics to improve understanding of the vast range of genetic resources that are available for this species.

“We have sequenced 200 ryegrass genomes, sampled from 10 separate populations distributed across southern and eastern Europe, and through analysing this dataset we want to build up a picture of the population structure to figure out the genetic relationships between them, going back many years. We really need a lot of storage just for our sequences even before we begin to analyse our data, and in fact that’s the first thing that Supercomputing Wales offered us,” Fu says.

The data analysis needed to find relationships between the plants is complex and time consuming, she says.

“You’ve got all these data and you need to clean them up, and then we start to call the SNPs, the single nucleotide polymorphisms.

“We use machine learning algorithms to predict the structure of the ryegrass genome first, and that requires a lot of predictions and recalculating round by round, and then we use Bayesian models to check whether we have found a real SNP or just a random sequencing error – is it meaningful or not?” says Fu.

“But luckily with Supercomputing Wales we can apply for multiple cores, say, 120 cores and work 120 times faster,” she says. “We just throw them into Supercomputing Wales and they run for you.”

The support team at Supercomputing Wales has always been helpful when it comes to installing and running the project’s software on the system, and fast to provide storage and compute time.

“You have to ask in advance for storage and time, but they are fast. They have a ticket system to request what you need, or ask about a problem, and there’s also a meeting every week with other users where we can discuss any issues and help one another,” Fu says.

Having worked with other HPC facilities internationally, Fu is pleased with how well Supercomputing Wales operates.

“It’s organised in a very good and efficient way. This one is the best for me!”

Simulating interactions in the first moments after the Big Bang

The Extreme QCD project, based at Swansea University, uses Supercomputing Wales’ infrastructure to simulate the interactions between quarks – the constituents of protons and neutrons – in extreme environments such as the high temperatures that existed in the first fractions of a second after the Big Bang.

“We’re looking at trillions of degrees Celsius. The universe was very different, and the quarks, and the force that binds them, behave differently,” says Professor Chris Allton of Swansea University.

The simulation of quarks requires large scale processing power and complex calculations due to both the strength of the force binding the quarks and to the fact that the force is quantum mechanical, forcing the inclusion of all possible motions of the quarks, no matter how seemingly unlikely. The Extreme QCD research team breaks the space it is studying into a lattice of points, each of which is analysed using complex equations, to create billions of variables.

Supercomputing Wales offers a “very clean, reliable system that’s very approachable,” Allton says.

“It’s a computing cluster that our code runs very naturally on, so it’s a very reliable, secure environment. It’s a nice scale, so we can so these enormous calculations, and it’s got the flexibility.”

While some Extreme QCD calculations are also run on other UK and European supercomputers, they can have speciality features which don’t fit the code as well or offer the same service as Supercomputing Wales, he says.

A close working relationship means that the Supercomputing Wales team is more likely to be able to let the project run code at short notice. Specialists from Supercomputing Wales often sit in on the project’s weekly collaboration meetings to provide advice.

The Supercomputing Wales team has also helped to adapt the Extreme QCD code, first to include vector optimisation to suit their own system and latterly giving advice about how to migrate the code so that it will run on the GPU-based High Performance Computing systems commonly used in other supercomputing centres.

“We’re lucky to have a lot of interaction with the team, and we do need the support as it’s becoming more and more specialised to write efficient high performance code. So apart from the actual machines, having the team to work with is fantastic,” Allton says.

While it would not be necessary for the Supercomputing Wales team to understand the science behind the research being done, many of the Research Software Engineers in the team come from a theoretical particle physics background, Allton says. That understanding of their work has been an added bonus to the researchers in their work with Supercomputing Wales.

Using Artificial Intelligence and data in manufacturing to maximise assets

The EPSRC-funded Transfer Learning for Robust, Reliable and Transferable Cyber Manufacturing Systems project, led by Cinzia Giannetti of Swansea University College of Engineering, works with industry partners to increase the use of artificial intelligence in manufacturing and help them use data to make better decisions.

“The main body of my research is looking at developing machine learning technologies, and algorithms that can use the vast amounts of data collected during manufacturing processes and use them to enable companies to make better, faster and optimal decisions,” says Giannetti.

Machine Learning can, for example, help a company to run what is called ‘predictive maintenance’, to maximise the use of assets and machines, Giannetti says.

“The idea there is to use data to develop and train very complex predictive models using deep learning and then use those models to predict the remaining useful life of equipment, and determine the right time to do maintenance.

“The research itself involves collecting and synthesising large volumes of data collected through different machines and sensors. We then want to visualise this data and use it to develop a predictive model,” she says.

“Given the amount of data, the precision we want to achieve and the complex architecture of Deep Neural Networks, we need an adequate computing facility,” Giannetti says.

The Research Software Engineers at Supercomputing Wales have been vital in getting this work done, she says.

“They provided training and, in the initial stages, helped us to perform some optimisation to run the models more efficiently. The biggest advantage of Supercomputing Wales is not only the availability of high performance hardware but actually having people that can support us in running the programmes – the training, the support, the help in setting up the system. We’ve become quite independent now, but at the beginning it was crucial to us,” Giannetti says.

Research across Wales supported by supercomputing facilities

Supercomputing Wales provides access to powerful computing facilities to high-profile science and innovation projects across Wales, with the aim of capturing more research funding, increasing scientific partnerships, creating highly-skilled research jobs and supporting collaborations with industrial and other partners.

The Wales Gene Park at Cardiff University takes advantage of the facilities, helping to advance its cutting-edge research that provides understanding, diagnosis and treatment of a wide-range of inherited diseases and cancer. At Swansea University, the facilities are used to generate global information needed for weather forecasting and improve models of the climate, with algorithms developed by the University used by the UK Met Office as part of its daily forecast. At Aberystwyth University, the facilities are used to support research projects including DNA sequencing for plant breeding, and the ‘Big Data’ challenges of earth observations, with the facilities used to analyse high-resolution satellite imagery to assess land cover and vegetation; whilst at Bangor University, the facilities support tidal energy and oceanographic projects, with opportunities for interaction with the ERDF funded SEACAMS 2 project.

Building the world’s fastest boat

Supercomputing Wales is contributing to the dream of building the world’s fastest boat.

Dr Ben Evans and the team of Supercomputing Wales Research Software Engineers at Swansea University are working with Norson Design to harness the power of supercomputing to develop faster, safer, more cost-effective vessels.

To address the design challenges presented by the complex behaviours of spray and turbulence, as well as an interaction surface between the boats which is dynamically-changing and a complicated shape, researchers at Swansea University applied methods used widely in aerodynamics. With HPC, the design team could use state-of-the-art techniques, like shape optimisation, and artificial intelligence techniques to efficiently generate AI-developed aero and hydrodynamic profiles for high-speed vessels.

“A typical simulation uses 450 compute cores running for 72 hours to get about 0.5 seconds of physically simulated time. It would take years on a desktop,” says Dr. Ed Bennett, Research Software Engineer at Swansea University.

“With supercomputing resources, it’s exciting to see advances in the marine industry. We are working on radical boat designs based on hydrodynamics and aerodynamics, carefully optimising the hull to help create the world’s fastest boat, and we’re very excited to see how it’s going to turn out.”

First ever detection of gravitational waves

At Cardiff University, the Gravitational Physics Group who in 2016 announced the first ever detection of gravitational waves as part of the LIGO (Laser Interferometer Gravitational-Wave Observatory) consortium will benefit from the upgraded Supercomputing Wales facilities. In the coming years, gravitational waves will allow researchers to peer into the cores of exploding stars and probe the structure of neutron stars, potentially revealing completely new and unexpected phenomena that will challenge our current understanding of the universe.


Get in touch

Supercomputing Wales
Data Innovation Research Institute
Cardiff University
Trevithick Building
CF24 3AA