32 new projects from Czech institutions awarded access to the LUMI supercomputer

In the 36th round of the IT4Innovations Open Access Grant Competition, 32 researchers and their teams from 11 Czech institutions have been granted access to the LUMI supercomputer.

Charles University and VSB – Technical University of Ostrava, including the IT4Innovations National Supercomputing Centre, were each awarded LUMI resources for six projects. Brno University of Technology and the Czech Academy of Sciences were each granted resources for five projects.

Most of the supported projects are single-year projects, meaning they can use the allocated resources until January 2027. In addition, five projects were awarded multi-year support, with access to the computational resources extending until January 2028.

Interest in LUMI resources among Czech researchers remains stable, reflecting the growing need for computing power in data-intensive science, advanced simulations, and artificial intelligence. At the same time, demand for IT4Innovations’ in-house supercomputing resources continues to exceed available capacity significantly. In particular, requests for Karolina CPU and Karolina GPU resources, which are, like the LUMI supercomputer, co-funded by the EuroHPC Joint Undertaking, reached almost double the number of hours available for allocation.

The LUMI-G partition will support researchers, particularly in artificial intelligence, machine learning, and GPU-intensive simulations, while the LUMI-C partition will be used for large-scale numerical modelling, data processing, and simulation tasks across a wide range of scientific disciplines.

Three selected research projects below are now using the LUMI supercomputer to tackle challenges spanning medicine to artificial intelligence.

Designing new peptidomimetics for immuno-oncology

At Charles University’s Faculty of Pharmacy in Hradec Králové, Jana Přecechtělová and Professor Pávek’s team are studying NR2F6, a nuclear receptor that plays a key role in the immune response. Their multidisciplinary team of experimental scientists and bioinformaticians uses molecular simulations and generative AI to design peptidomimetics – small protein-like molecules – that can disrupt NR2F6’s interaction with its co-regulator NSD1, potentially opening the door to new immuno-oncology therapies.

Peptides controlling membrane curvature

At CEITEC, Masaryk University, Robert Vácha’s team investigates how peptides and proteins interact with cell membranes, which are constantly changing shape. Some molecules recognise specific membrane curvatures, while others can actively alter the membrane shape. These processes are vital for substance transport, cell division, signaling, and viral infection. Using LUMI for large-scale molecular simulations, the team aims to design peptides that can detect or induce particular membrane curvatures, which could lead to new biotechnological and therapeutic tools.

Sign language recognition, translation, and interpretability

At the University of West Bohemia in Pilsen, Marek Hrúz and his colleagues develop systems for automatically recognising and translating sign language, including hand gestures, facial expressions, and body movements. Their AI system, SignLLaVa, converts signs into written text using visual recognition and a large language model. With newly granted LUMI resources, the team will refine the alignment between sign languages and their spoken counterparts, aiming to improve communication and access to education, employment, and public services for the deaf community.

Copyrights of the images belong to the respective research groups.