From planets to pandemics: translating science in challenging times
WATCH: Rohan fields some tough questions on lockdown compliance (at the 1:48 – 2:03 timestamp) for 7NEWS Melbourne on 20 September 2020. Image: 7NEWS Melbourne
After years of building NCRIS enabled software to address geodynamics problems, researcher and AuScoper from The University of Melbourne, Rohan Byrne, found an opportunity to turn his Everest code to help model COVID-19 mobility data during Melbourne’s 2020 and 2021 lockdown periods. Here, Rohan shares his story.
JC — Rohan, please tell us what you usually do as a planet builder?
RB — I use big computer simulations to study how the hot rocks deep beneath us shift and flow over billions of years. I’m trying to understand why planets that are born the same — like Earth and Venus — can go on to lead very different lives. And I want to apply that knowledge to the thousands of Earth-like planets we’ve discovered beyond our sun, to try to get a sense of whether ‘living’ planets like ours are common or rare in the universe.
It’s a big topic with many unknowns. In other fields with heaps of mysterious quantities, researchers try to cover their bases by running lots and lots of similar-but-different models. But I needed to run thousands of models, and no-one had done that before in my field.
The toolkit just wasn’t there, I had to create my own. The result was PlanetEngine, a ‘wrapper’ around AuScope’s Underworld geodynamic modelling code that makes it easier to orchestrate big, complex jobs.
“It slashed the labour required to do my research — but it wound up being the beginning of an even stranger journey.”
A snapshot of one of thousands of mantle convection models run by PlanetEngine so far; colour represents temperature, arrows are velocity, and contours show viscosity on a log scale. Image: Rohan Byrne
JC — What made you think of applying your code to new problems outside of geoscience?
RB — My eyes were really opened at the 2019 Australasian Leadership Computing Symposium (ALCS) led by the AuScope peer, the National Computational Infrastructure (NCI) in Canberra, 2019. Meeting so many wonderful people in diverse fields — from pharmacology to astrophysics — I was struck by the similarities of the challenges we faced.
“We have amazing resources in this country, but we struggle to make good use of them. And though we know we need to make our data FAIR, we can’t seem to find the hours in the day to do it. It’s part of what I call the ‘complexity crunch’, and I think it’s a big problem.”
I presented my work at ALCS and was totally overwhelmed by people’s positivity. There was a clear, strong appetite for creative and cross-disciplinary solutions. I took home best-in-stream and spent the summer ripping the guts out of PlanetEngine. That was the beginning of Everest: my attempt in code to tackle the ‘crunch’ for scientists of all stripes.
JC — How did you take the first leap into that journey?
RB — Well, we all know what happened next. The pandemic hit and I found I just couldn’t concentrate on my thesis anymore. I emailed the Doherty Institute and the next thing I knew, I was part of a cross-faculty team advising the government on pandemic control.
Australian cities had been slammed into lockdown but we had no way of seeing the effects. It became my job to turn an ocean of Facebook user location data into a daily-updated Crisis Mobility Data Portal that would reveal whether the policies were holding in real time.
Melbourne’s journey through 2020 shown in the Crisis Mobility Data Portal; the red line shows COVID-19 cases, while the blue line shows the ‘mobility score’, a measure of lockdown effectiveness. Image: Rohan Byrne
Using Everest, I was able to get our app up and running in six weeks flat. It remains today the only public-access resource for mobility tracking in this country.
JC — Four lockdowns later, how did you go?
RB — What started as a sprint became a marathon. Along the way, I met some great people - like Dr Jason Thompson from Melbourne’s Faculty of Architecture, Building and Planning, one of the social modellers behind Victoria’s “roadmap” out of lockdown of September 2020.
“Jason and I had both been involved in the scientific response to COVID-19 from the early days. Comparing notes, we quickly agreed that the pandemic had revealed a big problem: science is just too slow. The ‘complexity crunch’, once an inconvenience, had proved deadly in a crisis.”
That experience went on to inform the University of Melbourne’s entry into the Trinity Challenge, a new international coalition which aims to fortify societies against future health emergencies using novel data-driven approaches. Dubbed CrisisEngine, our contribution— a powerful new agent-based ‘social model’ built with Everest — is in part a continuation of the journey that began all those years ago with PlanetEngine. The University’s proposal was awarded an honourable mention in the Challenge’s highly competitive first funding round and is now moving into formal development.
A snapshot from the early development of CrisisEngine. Blue particles represent individual humans; red particles are active COVID-19 cases; yellow particles are recovered cases. Image: Rohan Byrne
JC — What response has your work evoked?
RB — I’ve been really humbled by the response. The Age, the Herald Sun, 3AW radio, and Channel 7 have all featured the Crisis Mobility Data Portal at different times, and thousands of citizens have shown their support in person and through social media.
“Our conviction from the beginning has been that seeing is believing: that data, when it’s open and visible, can change the world.”
I’ve had complete strangers thank me for helping them through the lockdowns — just being able to see the city’s response on a line graph has been proof enough for many that we really are ‘all in this together’.
JC — What excites you most about your experience?
RB — The pandemic has been a real shock to the system. There’s electricity in the air; things that concerned us before seem trivial. People joke about ‘dark times’, but the jokes aren’t funny anymore.
“The good news is that we have science on our side. The universities’ response to COVID-19 has outed a secret army of restless talent. Now it’s up to AuScope and all of our NCRIS colleagues to give that army an arsenal equal to the fight. Whether it’s CrisisEngine modelling disasters in real time, or the Downward Looking Telescope unearthing the materials for a sustainable future — if we build it, society will use it. That’s how we make a difference.”
JC — What would you like to see happen in the future to enable more transdisciplinary research like this? And, what’s next for you?
RB — For me, my immediate future is my thesis. I have to capture what I’ve learned, package the capability I’ve developed, and share it everywhere I can.
Beyond that, it’s clear that the academy can’t afford the luxury of disciplines anymore. At the University of Melbourne, Earth Sciences has just been merged with Geography: another crisis, another opportunity.
“But it’s communities of methodology like NCRIS, NCI, ARDC, AARNet, and Melbourne Connect that will have to shoulder most of the burden: providing the training, networking, advocacy, and — yes — the money that will make transformative, translational research possible.”
If we do this right, we will retire with pride in a much better world than this one.
STORY IN A NUTSHELL
AuScoper, Rohan Byrne has turned his NCRIS enabled coding skills from planet building to modelling COVID-19 mobility data in a collaboration with The University of Melbourne to assist in Victoria’s COVID-19 health response.
AUTHORS
An interview between Jo Condon
(JC) and Rohan Byrne (RB)
FURTHER READING
Crisis Mobility Data Portal
More cases expected from outbreak
in northern suburbs (The Age)