GPU-Cluster

• Subject Area: Basic Research in Biology and Medicine
• Term: Funded in 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 553375143

The exponential growth of genomic datasets, coupled with the increasing complexity of companion computational approaches, presents a critical bottleneck for computational biology research and its applications to human health. Our lab, dedicated to deciphering the genetic basis of gene regulation and its role in disease, urgently requires a significant expansion of our computational infrastructure to keep pace with these demands. As highlighted by our recent work pioneering the application of large-scale language models to DNA and developing deep learning-driven methods for de novo peptide sequencing, access to substantial GPU-accelerated computing power is paramount for our research. Therefore, this proposal seeks funding for a high-performance computational cluster comprising three GPU servers and a secure high-performance GPU server. Integrating with our existing resources, this new infrastructure will provide the computational power necessary to fully leverage large-scale datasets, such as the thousands of hematologic malignancy samples recently analyzed in our lab, and to develop and apply the next generation of deep learning models. This expansion is essential for making significant advances in our ongoing work on transformer models for DNA language modeling, gene expression prediction, and for significantly accelerating progress in optimizing variant interpretation algorithms – a critical step towards more accurate diagnosis and personalized treatment of genetic diseases. Furthermore, as active contributors to the German Human Genome-Phenome Archive (GHGA), we recognize the importance of secure and ethical data management. The secure high-performance GPU server, with its encrypted computing capabilities, will be instrumental in designing and prototyping algorithms for the secure analysis of sensitive health and individual genomic data, aligning perfectly with the stringent data governance principles of GHGA. This investment in state-of-the-art computational infrastructure is not only a technological upgrade; it is an investment in our ability to translate our research into tangible benefits for human health.

• DFG Programme: Major Research Instrumentation
• Major Instrumentation: GPU-Cluster
• Instrumentation Group: 7040 Vektorrechner
• Applicant Institution: Technische Universität München (TUM)
• Leader: Professor Dr. Julien Gagneur