MACSYS recently welcomed Dr Chris Rackauckas, Vice President of Modelling and Simulation at JuliaHub, for a multi-campus visit that spanned three of our five nodes: The University of Melbourne, UNSW, and QUT.  Over six days, Chris engaged deeply with MACSYS researchers – delivering seminars, leading technical workshops, and participating in wide-ranging one-on-one discussions.

As a MACSYS partner organisation, JuliaHub brings deep expertise in high-performance computing and scientific machine learning – capabilities that are central to MACSYS’ ambitious goal of building a computable whole-cell model.

“I think MACSYS is the most ambitious program I’ve seen that says, our goal is to build a whole cell model,” Chris said.

His expertise in scaling complex models brings a valuable perspective to MACSYS’ mission. He challenged researchers to think critically about what a whole-cell model should include and what it might leave out.

“I have this background in how to make models big,” he explained. “But the question becomes, what is the model that should be made big? What is it potentially including? What is it not including?”

The discussions also touched on the predictive power of biology, drawing parallels to physics.

“Can biology be understood at the same level as physics? Can we build models that are truly predictive,” he asked. “The math is in the right spot. The biology is in the right spot. The computers are in the right spot. This might be possible.”

His seminar, delivered at all three stops, focused on scientific machine learning and universal differential equations (UDEs), techniques that integrate neural networks into traditional scientific models to accelerate discovery.

“I really enjoyed the talks about scientific machine learning in Julia. I’m particularly interested to learn about some new computational tools for optimising stochastic systems. I had no idea this kind of analysis was even possible,” said Dr Tom Kimpson, a Postdoctoral Research Fellow at The University of Melbourne.

“Chris’ seminar was well received and also attracted applied mathematicians interested in numerical analysis of differential equation systems. It was interesting to discuss the possibilities of using neural network terms in the equations to suggest possible biological targets for exploration of specific effects in the dynamics,” said UNSW Chief Investigator Adelle Coster.

During his visit, Chris also highlighted the importance of integrating diverse modelling approaches and the role of modern tools such as machine learning, high-performance computing, and the Julia ecosystem in supporting MACSYS’ evolving needs.

“A lot of the people who do modelling like to talk with me because these tools were built for these large, difficult biological models in the first place,” Chris explained.

“He was passionate about sharing his knowledge, and it was exciting to hear about numerical methods and Julia programming from one of the world’s leading experts in the field,” said Dr Michael Pan, a Postdoctoral Research Fellow at UNSW.

Chris says the mission of the Centre and the ambition of its researchers has him optimistic about lies ahead.

“This has really been a nice overview of where the field is and where I can lend my expertise.”

MACSYS is excited to be working with Chris and JuliaHub, and we look forward to seeing the outcomes of the discussions that began with his visit.

(feature image:  Chris with QUT-MACSYS researchers)

Videos:

• VIDEO:  Chris discusses the prospects of modelling a whole cell
• VIDEO:  Chris talks about his MACSYS visit

Photos from Chris’ Visit: