European research is progressing at a fast pace in the domains of brain science and supercomputing. As neuroscientists need more and more access to high-end computing capacities, a new partnership will deepen the links between the two fields, paving the way for novel insights into basic brain function and to new approaches to treating and diagnosing brain disorders. Interestingly, new findings in neuroscience promise in return technological advancements that will increase the versatility and performance of supercomputers. Katrin Amunts, Scientific Director of the Human Brain Project and Jean-François Buggenhout, Head of the Future and Emerging Technologies (FET) Flagships Unit at the European Commission, share with you their views on the promising alliance of European Neuroscience and High Performance Computing.

The Human Brain Project, at the forefront of brain research

With the Human Brain Project (HBP), launched in 2013 as one of the two first major FET Flagship initiatives, Europe has become an international leader in advanced neuroscience research. In this field, understanding the functioning of the brain and its diseases requires to gather and process massive amounts of different types of data as well as running ever-complex models in simulation. This is why access to High Performance Computing (HPC) is becoming crucial for neuroscience. How to develop such an alliance was one central point of our discussions, last April, at the event "Shaping Europe’s Digital Future - High Performance Computing for Extreme Scale Scientific and Industrial Applications" co-organised in Sofia by the European Commission and the Bulgarian Presidency of the EU.

EuroHPC: the European push towards exascale computing

Fortunately, High Performance Computing is high at the European agenda, with the European High-Performance Computing Joint Undertaking (EuroHPC) proposed by the European Commission. It recommends the investment of EUR 1 billion into a world-class European supercomputer infrastructure, including two Exascale systems, i.e. systems able of performing 218 calculations per seconds, hundreds of times faster than the most powerful supercomputer of today.

Such systems are crucial for understanding the enormous multilevel complexity of the brain, resulting in extremely computationally challenging models and simulations of brain structure and dynamics.

The need for tailor-made systems

But for some research, the needs of neuroscientists can be quite specific. For example their work often requires interactive visualization and control. This is why HBP developed two prototype supercomputers JURON and JULIA as pilot systems for interactive supercomputing in 2016. Even with lower performance than the high-end supercomputers, they have proven their benefit for brain research. Researchers use these two systems to learn how to make best use of their architecture. This paved the way to a more ambitious project: HBP’s Interactive Computing E-Infrastructure (ICEI) project.

With ICEI, five European supercomputing centres – located in France, Germany, Italy, Spain and Switzerland –– are starting to build-up a powerful federated compute and data infrastructure called Fenix. The Fenix infrastructure will integrate a broad range of services, including scalable and interactive computing and various data hosting services. It will be an integral part of the Research Infrastructure that HBP is building up for neuroscience in Europe, which offers resources like the most comprehensive 3D atlas of the brain to date: the multiscale, multimodal HBP Brain Atlas. ICEI is tailored to support the goals of the HBP. This modular computing infrastructure will also be accessible to European researchers at large.

The ICEI project and the Fenix infrastructure enable the HBP researchers and neuroscientists to:

  • Integrate the enormous amounts of experimental data gathered by neuroscientists worldwide into a common framework, to produce the most comprehensive and detailed atlas of the human brain to date. This atlas is a freely available tool and resource to investigate all levels of the brain from the cell to large networks of brain areas.
  • Further develop brain simulation methods for simulation with ever larger detail and scale, speeding them up to biological real time, and provide this as a web-based service to the community.
  • Use innovative ICT to benefit patients suffering from brain disorders, as through the project EPINOV, short for “Improving EPilepsy surgery management and progNOsis using Virtual brain technology”. This clinical trial is developing brain models to improve epilepsy surgery. HPC is important to personalize the brain network models through the application of machine learning, to adapt them to individual patients.

A win-win situation of collaboration between Neuroscience and HPC

If the contribution of supercomputing to brain research is obvious, the benefits of neuroscience for HPC are equally important.

With their specific needs, neuroscientists help developing more versatile supercomputers based on innovative architectures, providing new capabilities such as visualization, interactivity and flexibility. This has the potential to bring HPC to new users, in fields such as material sciences and genomics.

But new insights into the brain’s astonishing complexity also drives progress in the core of HPC technology and shapes related fields like Artificial Intelligence. A better understanding of the functioning of the brain paves the way to neuromorphic computing, which promises to enhance traditional supercomputers with processors built on radically new approaches, based on interconnected artificial neurons or on digital representations of neurons. Neuromorphic computing will be key in developing the next generation of extreme-scale computing, i.e., modular supercomputing.

In practice, HBP neuroscientists are learning how to make full use of today’s massive computing capabilities. In turn, their needs inspire new supercomputing architectures and promise to enhance them with brain-inspired processors. We are proud to say that the Human Brain Project is an important use-case for the advancement of supercomputing and the science cloud in European research.