ESFRI Science Market
The HBP was recently very proud to host a Science Market for representatives from ESFRI, the European Strategy Forum for Research Infrastructures, at the Biotech Campus in Geneva. This was part of the 57th ESFRI Forum Meeting- The aim of the Market was to introduce the delegates to how the HBP is working towards creating a dedicated research infrastructure for understanding the brain.
In his opening remarks, Philippe Gillet emphasised the need for integrated infrastructures to explore and investigate complex scientific challenges. The HBP focuses on the creation of various Platforms (collections of software and hardware tools and applications) that can be used by neuroscientists, clinicians, computer scientists, mathematicians, theoreticians, engineers and roboticists, and researchers from various disciplines. Creating and bringing these together on this scale is a novel concept, and obviously invoves a very steep learning curve. Hopefully, however, this developing HBP Research Infrastructure will become part of the the overall scientific infrastructure for Europe, a process that has already begun successfully in other countries.
Karlheinz Meier (Universität Heidelberg, Germany) drew parallels between two of the most important questions facing modern science: how does the universe work, and how does the human brain work? Both are very complex systems involving multiple scales in space and time; one is based on physical laws, while the other is an internal model based on perception, action and learning. However, both require robust research infrastructures to explore them. Besides the appropriate instrumentation and ‘big data' analytics, computing power and simulations are increasingly used and are crucial to try to make sense of such complex systems. The world of astrophysics, for example, would not be where it is now without computing. The same could perhaps be said for modern neuroscience. The HBP is therefore building a research infrastructure with the same kind of elements used for investigating the mysteries of the universe. With this infrastructure coming together, combined with the rapid growth in knowledge from basic science and the subsequent theories being produced, these are indeed exciting times for the HBP and neuroscience.
In addition to the emerging infrastructure, other essential elements are collaboration, knowledge and innovation. Importantly, these need to be integrated so that the people who need to use them can do so quickly and effectively. For the HBP, the area where this integration takes place is the Collaboratory, as explained by Jeff Muller (EPFL, Switzerland). This is an online portal where interested parties can explore the HBP Platforms, work with web-based scientific tools to visualise and share data, collaborate with researchers from other areas, and discover how the tools can help expand their own work. Successful examples of this kind of integrated collaborative space already exist, but this is a relatively new concept for the neuroscientific community. The HBP Platforms (Neuroinformatics, Brain Simulation, High Performance Analytics and Computing, Medical Informatics, Neuromorphic Computing, and Neurorobotics) that can be accessed via the Collaboratory have been developed by over 100 Partners in 24 countries, and are physically hosted at computing centres throughout Europe. The way in which users are granted access depends on the particular Platform in question. The possibility to run different types of experiment and to interact with others therefore supports strong researcher collaborations and allows the development of reproducible digital science.
Neuromorphic computing, i.e. building computer systems inspired by the architecture and function of the brain, is a core part of the HBP. Dave Lester (University of Manchester, UK) explained how neuromorphic computing involves taking aspects of the principles of structure (e.g. cells, networks, an connections) and function (e.g. local processing, communication, and learning) in biology and transferring them to electronic circuits. Modelling neural circuits in this way will help to advance neuroscience – it will allow us to investigate learning, speed, robustness, and think about energy efficiency for computers. In the core project, the Neuromorphic Computing Platform offers access to two different and unique systems for modelling neural circuits: the SpiNNaker system (based in Manchester), a soft binary code model using microprocessors and which runs at real time, and the BrainScaleS system (based in Heidelberg), a physical custom mixed-signal model which runs at 10,000 times faster than real time. A third approach has arisen outside the HBP in the form of IBM's TrueNorth system, a fully digital system running on hard binary code. As such, there is no ‘correct' approach – all of the approaches so far have merit, depending on what the user requires, so complementarity between the approaches is important.
Building brain models is important, but it is only part of the story, since we also need to understand how the brain behaves in its natural habitat, i.e. a body, and how interactions between the two are controlled in the form of motor commands and sensory information. The relatively new science of neurorobotics, explained Marc-Oliver Gewaltig (EPFL, Switzerland), is therefore about giving brain models a body and investigating how the brain and body work together in dynamic environments. The Neurorobotics Platform, launched at the end of March, provides software and hardware tools for researchers so that they can perform experiments with brain models in different realistic virtual environments. This can be done in collaboration with other researchers, and can be used with brain models developed within or outside the HBP. Researchers can also interact with simulations of Roboy (http://roboy.devanthro.com), the soft-bodied humanoid robot that uses 3D printed bones artificial muscles and tendons. This and other features in the Neurorobotics Platform make robotics research easier, and more affordable and reproducible.
Clinical data about the brain and brain diseases resides in many different databases in hospitals, clinics and companies worldwide, but the ability to view relevant data from different places is currently extremely limited. Ferath Kherif (CHUV, Switzerland) introduced the Medical Informatics Platform as a framework for bringing together disparate clinical data, anonymised to ensure patient privacy, to make it possible to identify biological signatures of brain disease that are based on biology rather than symptoms. By combining and examining data from different sources, researchers can more easily look for patterns in the data that were not apparent before, and which may lead to different perspectives on certain diseases. Machine learning algorithms based on these signatures can then be developed. In this way, much more data will become available, and one of the challenges is to to find new and smarter ways of using it effectively. This is, of course, a highly collaborative project that will involve clinical and life science researchers, doctors, software developers and the general public, for the transfer of knowledge and integration into patient care.
Stephan Eilemann (EPFL, Switzerland) introduced interactive supercomputing, one of the tasks of the High Performance Analytics and Computing Platform. Supercomputers have traditionally been used in fairly conventional ways, but recent advances are resulting in a major technology shift in the way that scientists will use them in the future. These changes are not specific to the HBP, but are also impacting other domains. The use of supercomputers in the scientific field will substantially reduce iteration times in the scientific process and enable new use cases that were not possible using classical approaches. For example, supercomputers will enable scientists to explore circuit structure and connectivity and see how certain cells are selected. The aim is to move high performance computing towards interactive supercomputing, which will have different applications depending on the Subprojects. The impacts on the Project will include a reusable, open source system, scalable volume rendering, and neuroscience local field computation. These approaches will have a particular bearing on neurorobotics and brain simulation.
The HBP's path towards a Research Infrastructure, to be developed in the next phase of the Project, was elucidated by Jeff Muller. In addition to the base infrastructure, which includes national computing centres, cloud resources and networking, key components for the Research Infrastructure are the HBP Core Project ICT Platforms and the Collaboratory, which acts as the main entry point for the Platforms. The research infrastructure components from the HBP Core Project will combine with existing national infrastructure components to create an integrated HBP Research Infrastructure. A new legal entity for the HBP will also play an important role in coordination, working with funders, and supporting the marketing and user engagement strategies, allowing the HBP Research Infrastructure partners to focus on the scientific and technical work necessary. It will also help bring the emerging technologies to Technology Readiness Level 6 (technology demonstrated in a relevant environment). This will allow strong user communities to be defined and built before expansion of the infrastructure, and will allow the core project participants to follow a well-defined path to infrastructure funding. It is important to note that not all of the Platforms are at the same stage of maturity or have the same kind of base infrastructure requirements. However, Plartforms can be moved through the infrastructure pipeline independently if necessary, and integrated into the HBP Research Infrastructure once sufficiently mature. Over the next 18 months of the Project, the crucial steps will be to identify the relevant base infrastructure needs and the service levels required, create a legal framework, and form consortia to bring together the relevant partners that own and offer the necessary components for the HBP Research Infrastructure.
In closing, Karlheinz Meier noted that, while the focus may be on infrastructure, the HBP is cncenred with much more than this. Collaboration is key to the success of the Project;, the HBP is therefore actively building an interdisciplinary community. Education and training are also crucial, since success depends not only on the researchers of today but on the students who will be the researchers of the future. The audience was then invited to take part in various interactive sessions demonstrating some of the functionalities and features discussed, uncluding how the Collaboratory acts as the gateway to the Platforms, how the medical Informatics Platform will help to understand and treat brain diseases, how neuromorphic computing is making machines that learn, how data can be applied and used in the Neuroinformatics Platform, and how Roboy can help users of the Neurorobotics Platform.
What People are Saying
Collaborate, collaborate, collaborate. This is our opportunity.
The Human Brain is the most complex system that we know of. We would like to develop some kind of ‘google' brain where we can zoom in and out, see it from different perspectives and understand how brain structure and function is related. The ultimate aim of the Human Brain Project is to understand the human brain. This is only possible when we understand the structural organization of the human brain.
The Human Brain Project will become a major driver of ICT in Europe.