Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen (GWDG)

We operate several high-performance computers and conduct our own research on them in the field of secure and highly available computing infrastructure and efficient AI applications. We also participate in national (KISSKI) and Europe-wide (HammerHAI and HPC SPECTRA) initiatives for the standardization and improvement of HPC and AI applications and the improvement of user training. All research work on generative AI, data protection and storage solutions also forms an important basis for services that are made available to the MPG and other users.

The interdisciplinary and medical sub-project of ABA, funded by the Volkswagen Foundation, brings together experts from various fields for colon cancer research in order to stratify patients for an optimal choice of therapy. Samples of intestinal tumors are examined in 3D using imaging methods at the University Medical Center Göttingen (UMG) and the Max Planck Institute for Multidisciplinary Natural Sciences (MPI-NAT) and stored in a data lake at the GWDG. The data is compressed and analysed using AI on the high-performance computer Emmy operated by the GWDG. 

Computing power and storage capacity are of increasing importance in a growing number of scientific disciplines, including and especially in the life sciences. Due to advances in biotechnology and the successively increasing amount of data generated in this field, a highly efficient computing infrastructure for implementing these very methods is becoming increasingly important, in addition to methods of artificial intelligence (AI) and machine learning.

Data lakes have experienced increasing popularity for years and are being used in more and more institutions as central storage for all accumulating data, especially unstructured data. An unique selling point of a data lake is that data is stored in the respective raw format. This is intended to prevent information loss and thus increase the reusability of the data. Furthermore, a data lake at the GWDG provides means for data management and, thus, can also support researchers in complying with good scientific practice. 

The EU-funded project ”Open Forecast” is developing a generic platform for processing open, freely available data implementing automated data analysis pipelines and integrating high-performance computing ressources. Two use cases based on this platform are realised: smart cities and smart farming. In the first case, a complex meteorological 3D model is used to calculate forecasts of particulate matter pollution for the greater Stuttgart area. In the second case, forecasts for agriculture are generated based on satellite data from the Copernicus earth observation program.