Simulating the Human
In the wake of the COVID-19 pandemic, SuperMUC-NG is helping researchers fight coronavirus. The supercomputer at the Leibniz Supercomputing Centre (LRZ) in Garching is currently busy filtering out active compounds from drug libraries and millions of chemical substances, natural remedies, or approved drugs that can be used to defeat the coronavirus, or SARS-CoV-2. "Supercomputers are a remarkable resource for the development of COVID-19 treatments,” Peter Coveney, Director of the Centre for Computational Science (CCS) at the University College London (UCL), says. Coveney has extended experience working in this realm, and served as lead on the EU-funded projects CompBioMed and CompBioMed 2. “These machines help us identify possible treatments through a variety of ways, including machine learning, complex molecular dynamics, and artificial intelligence methods. Not only do we need to find molecules that bind to the spikes on the coronavirus, but we also need to model how well these bind when we know the spikes move around."
Digital twins for personalized medicine
The coronavirus crisis underscores the societal benefit of CompBioMed. Over the last five years, scientists and universities from Europe and the USA, under Coveney's leadership, have developed around 20 high-performance computing (HPC) programs, algorithms, and applications that can be used to create a virtual human model. These programs can be used to calculate and simulate the movement of muscles and bones, biochemical and molecular processes, as well as phenomena in the circulatory system. The codes and programs process big data sets to create a "virtual human being." This essentially means creating a “digital twin,” or virtual representation of a dynamic, real-world process—such as the myriad functions happening inside the human body. These digital twins, when infused with quality data, serve as a starting point for personalized medicine.
In the fight against COVID-19, CompBioMed's tools are now paying off: the simulation programs help to model the coronavirus and its reproduction behavior, test molecules for their efficacy, and discover antidotes as quickly as possible. Thanks to HPC, detailed models of the coronavirus were available only weeks after the outbreak of COVID-19, and HPC helped quickly identify the first 40 active substances with which the virus could be fought.
Human model in HD resolution
CompBioMed’s successor project, CompBioMed 2, started in 2019. While LRZ has supported the project since 2016, it became a core partner in 2019, and the organization is responsible for the analysis and management of large amounts of data.
In the second phase of the project, researchers will use supercomputers and artificial intelligence workflows to visualize three-dimensional, true-to-scale models of the human body in high resolution. The first milestones have already been reached: with data from the Swiss IT'IS Foundation and the CompBioMed blood flow program HemeLB, researchers used SuperMUC-NG to visualize the flow of red blood cells in the veins. "Compared to an expedition to Mount Everest, we are now moving towards Camp 2," Coveney said when describing the status of CompBioMed. HemeLB was the first application to use all 311,040 compute nodes of SuperMUC-NG.
Data management support
In addition to HPC consulting, training and support , LRZ will also help process ComBioMed data and catalogue it with metadata. This helps researchers more easily access relevant data sets wonline. The LRZ has already set up the storage system, LRZ Data Science Storage, for similar tasks, and offers tools for automated high-throughput analysis of huge amounts of data and for data security.
In CompBioMed 2, LRZ will also cooperate with EUDAT, the Open European Science Cloud. "This is interesting development, because EUDAT offers a rich toolbox for data management, which we might be able to combine with our services," says Stephan Hachinger, head of the Research Data Management team at LRZ. CompBioMed 2 can benefit from the experience that the LRZ team is already gathering in data management, projects such as LEXIS or Generic Research Data Infrastructure (GerDI), and vice versa.
Smarter through artificial intelligence
CompBioMed 2 also incorporates big data and artificial intelligence specialists at the LRZ. "Machine learning has arrived in research and medicine; it is used everywhere and complements classical computer simulation," says Peter Zinterhof, member of the Big Data and Artificial Intelligence team at LRZ, "By examining the functioning of trained networks, we can learn more about the inner workings of the problem." Zinterhof, who is also in charge of the DigiMed project on digitisation in Bavarian medicine at the LRZ, expects that artificial intelligence and machine learning, image recognition and other smart systems will therefore become more widespread in medicine.