LATEST RESEARCH RESULTS

Deep learning is revolutionizing protein science, with graph neural networks (GNNs) and multimodal models enabling unprecedented insights into protein function and design. In this project, the team led by Prof. Dr. Holger Gohlke developed two complementary AI models: TopEC and OneProt. TopEC uses 3D GNNs to predict enzyme functions directly from protein structures, incorporating atomic distances and angles to achieve high accuracy across more than 800 enzyme classes. Its structure-aware approach outperforms traditional 2D methods and remains robust even when binding site information is uncertain. In parallel, OneProt extends the multimodal ImageBind framework to proteins, aligning structural, sequence, text, and binding data into a shared…

Plants may not have brains, but they do have glutamate receptor–like proteins (GLRs) that behave much like the ion channels driving learning and memory in animals. These mysterious channels regulate key plant functions, from nitrogen use to pollen growth, yet how they control ion flow has remained unclear. To uncover their secrets, researchers led by Prof. Dr. Holger Gohlke combined high-precision modeling and molecular dynamics simulations of moss GLRs. Using AlphaFold2, they built accurate 3D channel structures and revealed how subtle changes in pore residues reshape ion permeability. The mutant channel showed enhanced calcium flow, linked to a more electronegative pore, an insight confirmed by large-scale simulations on supercomputers.…

The influence of compressibility effects on wall bounded flows is still not fully understood, especially when investigating its interplay with methodologies of drag reduction in engineering type flows. This project dealt with the application of oscillation control to supersonic turbulent channel flow. This method, well investigated for incompressible flow, was analyzed with respect to the influence of compressibility on the control effectiveness, by varying Reynolds and Mach numbers or adding tailored dissipation terms, to separate the effect of intrinsic and variable property compressibility effects. Additionally, the flow control was seen to strengthen the effect of the so-called very large anisotropic scales (VLAS).

Neutron stars are the most compact objects in the Universe with typically 1.5 times the mass of our Sun compressed into a sphere of just about 25 km in diameter, implying central densities higher than those in atomic nuclei. Most neutron stars are formed as remnants of massive stars when the degenerate core of these stars becomes gravitationally unstable and collapses, while most of the stellar matter is ejected in a violent supernova explosion with velocities up to 10,000 km/s. Two such neutron stars in a binary system can collide in a violent merger event after having approached each other on a spiral orbit over hundred of millions to billions of years, driven by the continuous emission of gravitational waves.

Stars are cosmic fusion reactors, which gain energy by nuclear reactions of light atomic nuclei to heavier ones. In stars of more than about nine solar masses, a sequence of burning phases thus assembles successively heavier chemical elements, starting from hydrogen fusion to helium as in the Sun, and continuing with helium, carbon, neon, oxygen, and silicon burning until a core of iron builds up at the center of the star. Iron as the atomic nucleus with the highest binding energy per nucleon cannot produce energy by further burning, and thus the growing iron core cannot escape a catastrophic end.

The strong interactions as part of the Standard Model of particle physics are described by Quantum Chromo-dynamics (QCD). Due to its strong coupling at typical energy scales in today’s Universe, predictions for strongly interacting matter, such as the one of the quark-gluon plasma, appearing in collisions of heavy nuclei at the Relativistic Heavy Ion Collider (RHIC) and future efforts, cannot be obtained using perturbative methods. The numerical treatment of QCD, discretized on a spacetime lattice – lattice QCD – has proven to be a viable tool to investigate the properties of QCD in the strongly coupled regime.

Compact objects, such as black holes and neutron stars, emit gravitational waves – tiny ripples in the fabric of spacetime – when they orbit around each other and eventually merge. The era of gravitational-wave astronomy began with their first direct detection of a binary black hole merger in September 2015. Just two years later, the first simultaneous detection of gravitational waves and electromagnetic signals generated by the merger of a binary neutron star system has been made. This multi-messenger event provided unique insights into the physics of compact binary systems, allowed for the testing of theoretical models for the emitted gravitational and electromagnetic waves, and enabled studies covering subatomic to cosmic length scales.

Intensified production has led to remarkable gains in chicken production (5x), growth rate (3x), and milk yield (2x). However, this relies on narrowed genetic diversity in primarily European commercial breeds, increasing their vulnerability to future biotic (disease) and abiotic (climate) stresses. Local breeds, critical for adaptation, are experiencing alarming declines (34% endangered, 5% extinct in 15 years) [1]. This is driven by the adoption of high-yielding European breeds, particularly in Africa. While initially productive, these breeds are often poorly adapted to local environments and lack disease resistance. Our research leverages genomics to understand the impact of industrialization and globalization on livestock genetics.

On Earth, the water cycle proceeds by three processes. First, water evaporates from the ocean and lakes. Second, water vapor condenses to form clouds. Third, clouds precipitate droplets of water which fall to the Earth as rain. After the rain falls to the Earth, the water droplets collect and eventually flow to lakes and oceans to be evaporated again. Galaxies cycle their star-forming fuel (called “baryons” in distinction to dark matter) in a similar three-step process.

Globular clusters (GCs) are massive and ancient star clusters populating practically all present-day galaxies. While GCs are frequently found in the galactic outskirts, the central regions of galaxies are occupied by more massive, aptly named nuclear star clusters (NSCs), and massive black holes (MBHs). Due to their ubiquity, GCs, NSCs and MBHs are thought to originate as natural by-products of the extreme gaseous and stellar densities that occur during the assembly of galaxies. The seeds of MBHs may have formed through runaway collisions in dense proto-GCs, and the spatial coexistence of NSCs and MBHs suggests a common mass-growth scenario.

Quantum Chromodynamics (QCD) is without any reasonable doubt the correct fundamental theory of the strong interactions between quarks and gluons. Thus, it describes exactly all properties of particles like the proton, which is a bound state of quarks and gluons, so-called hadrons. However, while it is rather straight­forward to calculate all properties of atoms, i.e. bound states of electrons and nuclei, from Quantum Electrodynamics, the theory which describes the electromagnetic force, this is not the case for QCD. In fact, the QCD­equations are so difficult to solve that many or even most properties of hadrons are only poorly understood at the QCD level.

Biologics offer significant therapeutic advancements, but their complex nature poses formulation challenges due to their sensitivity to environmental factors. Molecular dynamics (MD) simulations can accelerate formulation development by investigating biophysical behavior and guiding the design of stable and effective formulations. However, MD simulations face limitations in computation-al resources and force field accuracy. This report details the use of supercomputing capabilities for large-scale MD simulations, exploring various formulation parameters using coarse-grained (CG) models.

The idea of a thin, weak layer beneath the Earth’s rigid outer shell is over a century old, long predating the widespread acceptance of plate tectonics. This layer, termed the asthenosphere, was originally proposed in order to account for Earth’s internal support of topographic loads such as mountain ranges (isostasy). Our view of this layer expanded following the plate tectonic revolution of the late 1960’s, after which it came to be viewed as a lubricating layer to facilitate plate motion; passively flowing in response to the overlying plates. In the past two decades, it has become clear that the asthenosphere is a much more exciting and active part of the convecting mantle than previously imagined.

Invented and patented by Nikola Tesla in 1913 [1], the Tesla turbine is a unique type of turbomachine that features a rotor made from stacked disks with small gaps in between, as shown in Figure 1. It is driven by a fluid that enters the disk gaps at the outer circumference and travels on a spiral path radially inward towards central exit openings. Figure 2 visualizes the path of the fluid. Rotor torque is generated purely from friction between the working fluid and the rotor disk surfaces.

To make Tesla turbines more accessible to engineers, researchers in the past developed and validated methods for predicting turbine performance mathematically.

The scientific breakthrough associated to the LIGO-Virgo observation of gravitational waves (GWs) and electro-magnetic (EM) counterparts from a binary neutron star merger (BNSM) has been crucially supported by theoretical predictions provided by simulations in numerical general relativity (NR). Simulating the spacetime and the neutron-star matter fields in 3 spatial dimensions (plus time) is the only way to connect the strong-field dynamics to the observable gravitational and electromagnetic spectra. Crucially, these HPC simulations provide precise calculations for the GWs and for mass outflows of neutron rich material. The former are necessary to detect the signals and identify the properties of the source (masses).

Intracranial aneurysms, affecting approximately 3% of the western population, pose a serious threat due to the risk of rupture, leading to irreversible disabilities or death. Leveraging the computational power of modern supercomputers, our project employs the lattice Boltzmann method (LBM) to delve into the complexities of hemodynamics within intracranial aneurysms, utilizing our in-house numerical solver, ALBORZ.

Our research addresses the challenge of complex geometry in patient-specific aneurysms by introducing a curved boundary condition, enhancing the accuracy of LBM simulations.

Stratocumuli are low level clouds at the top of the atmospheric boundary layer, at altitudes of about 1 km and with thicknesses of about 100 m. They are key elements of the climate system. On the one hand, they extend over thousands of kilometers in the eastern boundaries of the subtropical oceans, e.g., off the coasts of California, Peru, and Namibia, favored by the temperature contrast between the cold upwelling water and the warm subsiding air. On the other hand, they reflect more incoming solar radiation (higher albedo) than the underlying surface of the ocean, while they emit similarly in the long-wave range. This combination of large coverage and net cooling effect substantially affects the Earth's radiative budget.

Combustion in most engineering applications, such as spark ignition engines and gas turbines, often involves elevated pressure conditions and non-unity Lewis number Le fuel-blends. However, under these extreme conditions, the flame morphology becomes increasingly complex and turbulent, persistent with convoluted structures arising due to the presence and interactions of inherent flame instabilities [1]. In this project, direct numerical simulation (DNS) analysis is performed to evaluate and complement existing modelling approaches to account for realistic operating conditions for combustion applications.

Over the last few decades, ab-initio methods such as density-functional theory have become sufficiently accurate to allow for the prediction of many properties of new crystal structures. However, these predictions come at a significant cost, and due to the vastness of the space of possible materials, theoretical material discovery remains one of the most challenging questions in materials science. Machine learning methods, trained on existing databases of ab-initio calculations, have the potential to massively accelerate the process of theoretical materials discovery. One of the most important properties targeted is thermodynamic stability, which is used as a proxy to estimate the probability that a given compound can be synthesized.

The VIRTUALLAB project made use of the JUWELS cluster to perform high-resolution atmospheric simulations based on measurements from various meteorological sites and field campaigns. The resolution of these numerical experiments is high enough to resolve small-scale phenomena such as turbulence, convection and associated clouds. These simulations create a "virtual laboratory", allowing scientists to fill existing data gaps to increase our insight into these phenomena and further improve their representation in models used for weather- and climate prediction. Various climate regimes were simulated including the marine subtropics, mid-latitude continental areas, and the high Arctic.