Awarded DECI Projects – DECI 8

Find below the results of DECI-8 (Distributed European Computing Initiative) call.

Projects were from the following research areas:

Astro Sciences Projects

Project Title: Towards an initial mass function of planetesimals

Project Leader: Dr. Anders Johansen, Lund University, Astronomy and Theoretical Physics, Lund, Sweden

Resource Awarded: 6 200 054 core hours on FZJ – JuRoPA, ICHEC – Stokes and RZG – RZG P6


  • Dr. Ross Church – Lund University, Astronomy and Theoretical Physics, Lund, Sweden
  • Prof. Melvyn B. Davies – Lund University, Astronomy and Theoretical Physics, Lund, Sweden
  • PhD Student Michiel Lambrechts – Lund University, Astronomy and Theoretical Physics, Lund, Sweden
  • Dr. Mordecai-Mark Mac Low – American Museum of Natural History, Astrophysics, New York, USA

The planets of the solar system and the exoplanets – planets that orbit stars other than the sun – are a fascinating research area. Fuelled by new detection methods that find more and more planets around other stars, satellite missions to other planets, moons and asteroids in our solar system, and the ever growing power of supercomputers, planet research is in rapid development and enjoys lots of interest from the broad public. The field is closely tied with the perhaps most fundamental question of all: how common is life in the universe? – a question pursued by astronomers, geologists, biologists, physicists and chemists in concert. Planets form in discs of gas and dust orbiting young stars as dust and ice particles collide and grow to ever larger bodies. An important stage in the planet formation process is the formation of km-scale planetesimals. Planetesimals are building blocks of both terrestrial planets like the Earth and of the solid cores of gas giant planets such as Jupiter. A fundamental problem is that cm-sized pebbles do not stick when they collide. Supercomputer simulations performed by members of our group have identified a surprising phenomenon that allows growth from pebbles to planetesimals: pebble-sized particles concentrate in dense filaments that protect them from gas drag, in a process related to why bicycle riders and migrating geese travel in groups. The densities of pebbles get so high that gravity takes over and leads to gravitational collapse to form planetesimals. The aim of this research project is to use high-resolution computer simulations to understand the birth sizes of planetesimals. The asteroid belt between Mars and Jupiter and the Kuiper belt beyond Neptune are examples of planetesimal belts left over from the planet formation process. The largest asteroids and Kuiper belt objects have sizes that are similar to the largest planetesimals that form in the computer simulations, but an important feature of both these populations is that the size distribution of the planetesimals show a break around 50 km in radius. This has been dubbed the missing intermediate-sized planetesimals problem. Previously we have in our computer simulations only been able to form the largest planetesimals (with radii of 150-1500 km) from overdense filaments of pebbles. Small planetesimals form from small-scale particle overdensities and hence it requires very high resolution simulations to model their formation. Modelling planetesimal formation at much higher resolution than previously, using PRACE supercomputers, we will investigate the size distribution of planetesimals down to 30 km in radius and compare to the observed size distributions of asteroids and Kuiper belt objects. We will also monitor the fraction of newly born planetesimals that are in binaries and compare to the binary fraction of the Kuiper belt. This will give us insight into the dominant physical processes that govern planetesimal formation and hence give constraints on how efficiently planetesimals grow to form planets.

Bio Sciences Projects

Project Title: Chlorite degradation by chlorite dismutase enzymes

Project Leader: Pietro Vidossich, Universitat Autònoma de Barcelona, Unitat de Química Física, Barcelona, Spain

Resource Awarded: 3 240 000 core hours on CINES – JADE-Harpertown and UHEM – UYBHM

Chlorite dismutase (Cld) is a unique heme enzyme able to decompose chlorite (ClO2-) into chloride and molecular oxygen (ClO2- → Cl- + O2, reaction 1). The enzyme shows high specificity for reaction 1, in contrast to other heme enzymes which are known to react with ClO2- though not generating O2. Our objective is to determine the reaction mechanism of reaction 1 as catalyzed by Cld and pinpoint the structural motifs at the origin of the enzyme striking specificity. To this scope, we will use ab initio (Car-Parrinello) and classical molecular dynamics simulations to determine the free energy profile of reaction 1 as well as of ligand binding and intermediate(s) dissociation. Free energy simulations of biochemical reactions are a formidale task, given the requirement of employing realistic model systems to account for the confining and electrostatic effects of the protein environment and the need of extensive sampling of conformational space in order to account for finite temperature effects. Reaction 1 is highly beneficial for detoxification of waste waters from industrial plants where oxochlorites are used. Unfortunately, the lifetime of Cld is limited under high substrate concentrations, hampering its use in bioremediation applications. Eventually, the insight gained from simulations may lead to the engineering of improved biocatalysts or inspire the design of a biomimetic catalyst for wastewater treatment.

Project Title: Free energy studies of cytochrome bc1 interactions with ubiquinone and ubiquinol

Project Leader: Tomasz Rog, Tampere University of Technology, Finland

Resource Awarded: 4 200 000 core hours on PDC – Lindgren


  • Artur Osyczka – Jagiellonian University, Krakow, Poland

The cytochrome (cyt) bc1 is an electron transfer complex situated on either the inner mitochondrial membrane of eukaryotes, or the plasma membrane of bacteria. It is an important redox carrier of the multi-phase electron transport chain that up keeps major part of our body’s energy metabolism. The reaction mechanism of cyt bc1 complex, also referred to as the Q- cycle transfers electrons to the cyt b subunit to generate a proton gradient across the mitochondrial membrane. During one cycle two hydrogens move into the mitochondrial matrix or negative (N) side and four protons move into the positive (P) side or the inter membrane space, or periplasmic side, in prokaryotes. Ultimately, the hydrogens pushed against the electrochemical gradient flow back to the mitochondrial matrix; the influx is coupled to generate adenosine triphosphate (ATP) molecules by ATP synthase.

In spite of its importance, the detailed mechanism of the proton and electron transfer in cyt complexes has remained unknown. This is mainly due to limitations of the current experimental techniques. The main sources of knowledge about cytochrome are studies of mutants that allow one to determine the importance of each residue. However, all these studies are indirect, and the proposed views have been debated. In this study we will perform molecular dynamics simulation studies of the cyt bc1 complex in explicit lipid bilayer, at physiological salt concentration. The main task of this study will be to calculate the free energy of substrate binding, and to elucidate substrate bonding modes.

Project Title: Transport Mechanism inn Drug Efflux Proteins

Project Leader: Prof. Simon Berneche, University of Basel, BIOZENTRUM, Basel, Switzerland

Resource Awarded: 1 450 000 core hours on CSCS – Rosa


  • Prof. Anna Seelig – University of Basel, BIOZENTRUM, Basel, Switzerland

Drug absorption across biological membranes is determined by passive influx and active efflux of drugs by trans-membrane proteins of the ATP binding cassette (ABC) transporter family, which notably includes P-glycoprotein (P-gp). The primary function of P-gp, which is well conserved across species, is to protect organisms against toxic compounds. At the same time, such natural property of Pgp gives rise to a phenomenon called “multi-drug resistance” which implies that drug cannot reach their target because they are recognized as foreign molecules. It causes major problem in cancer therapy and in antibiotic therapy (bacterial ABC transporter function in a very similar way). High resolutions structures of P-gp and the homologous bacterial Sav1866 have been recently solved and gave new impetus to the study of drug-resistance mechanisms. However, while a plethora of functional and structural studies provide a general view of the mechanism, the underlying microscopic details remain elusive. Like for many membrane transport proteins, the function of P-gp relies on three elementary mechanisms: selectivity, permeation and gating. Using molecular dynamics simulations we aim at identifying the key molecular interactions that sustain these basic functions. We will perform simulations of both the P-gp and Sav1866 proteins embedded in a lipid membrane. The structures of both proteins reveal a cavity at the level of the membrane through which substrates are recruited. How the proteins interact with its lipid environment is not clear, e.g. the number of lipids potentially occupying the cavity is unknown. The first stage of our project will thus involve a series of test simulations to identify the optimal lipid configurations. Next, our focus will be placed on the mechanism allowing the substrates to move from the lipid membrane to the core of the transporters. Based on a comparison of the chemical properties of known P- glycoprotein substrates, conserved patterns of hydrogen bond acceptors were recognized as a determinant feature of selectivity (A.Seelig, Eur.J. Biochem. 1998). Using different molecular mechanics approaches that are less computationally demanding, we have already identified some residues that could potentially serve as H-bond donors. Our goal will now be to verify some of these hypotheses using all-atom simulations, which is the only approach that can reliably describe the fluctuations of the molecular systems and notably the role played by water molecules in facilitating or hindering the binding of substrates. Finally, we would like to understand how the binding of ATP to the nucleotide binding domains favours conformational changes that leads to the efflux of drugs or other substrates. We have already observed in P-gp simulations that the nucleotide binding domains can undergo large displacement and come into contact with each other even in absence of ATP. Our hypothesis to be verified is that the binding of ATP stabilizes the contact between the two binding domains. Presence of lipids in the cavity could influence this mechanism. By providing a microscopic description of the transport mechanism in drug efflux pumps, we hope to contribute to a better understanding of issues related to drug-delivery and drug-resistance.

Project Title: Accurate calculation of drug binding thermodynamics and kinetics

Project Leader: Dr. Francesco Luigi Gervasio, Spanish National Cancer Research Centre, Madrid, Spain

Resource Awarded: 4 648 000 core hours on EPCC – HeCToR XE6


  • Dr. Giorgio Saladino – Spanish National Cancer Research Centre, Madrid, Spain

Notwithstanding a wealth of new genetic data, the treatment of serious and complex conditions, as cancer, is progressing slower than anticipated. The huge and ever increasing investments in R&D of new chemical and biological entities fails to translate into new approved drugs [1][2]. One of the reasons for this problematic situation is that many compounds in the drug discovery pipeline fail in late stages due to lack of efficacy and toxicity. To address this problem, many computational methods are being developed that try to predict these parameters in early development stages [3]. Unfortunately, despite significant recent developments, fast but approximated approaches still fail to quantitatively predict the thermodynamics and kinetics of drug binding to their targets, especially in the case of flexible targets. This is in large part due to the complexity of a quantitative estimate of the free energy profile and kinetics of binding, which often involves a description of complex target conformational changes, of the enthalpy/entropy compensation and of the hydration/dehydration effects. Fully atomistic molecular dynamics (MD) simulations coupled to the recently improved force-fields and novel advanced sampling techniques [4-6] have recently reached the time-scales and precision needed to explore the association of a drug to its target in a realistic way [7]. The time seems to be ripe to simulate complex cases of molecular association, for which experimental data is available or can be measured, and quantify the role of conformational changes (see e.g. [8]), of hydration/dehydration and entropy/enthalpy compensation in great details. In the last two years we devised an integrated approach based on the combination of free energy methods [4] path-like collective variables [5] and transition path sampling [6]. The approach is computationally expensive, but it is much more efficient than brute-force MD simulations. It allows the reconstruction of the thermodynamics and kinetics of drug binding within the limit of the force-field. We will apply it to study the association of drugs and drug-like molecules to two classes of very flexible and pharmaceutically important anticancer targets, namely the Abl and Src tyrosine kinases and the heat shock protein 90. Our state-of-the-art simulations will be fully validated by micro-calorimetry, surface-plasmon resonance experiments and NMR. The experimentally validated results will provide a solid basis for the understanding of subtle biophysical effects in molecular association and guidance for the systematic improvement of the target and drug force-fields, when needed. [1] Walker,S.M. and Davies,B.J. Drug Disc. Today 16 (11-12), 467–471 (2011) [2] Bunnage,M.E. Nat. Chem. Biol. 7(6), 335–9 (2011) [3] Van Drie,J.H. et al. J. Comput-Aided Mol. Des. 21 (10-11), 591–601 (2007) [4] Laio,A. and Parrinello,M. Proc. Natl. Acad. Sci. USA 99 (20), 12562 (2002) [5] Branduardi,D., Gervasio,F.L. and Parrinello,M. J. Chem. Phys. 126, 054103 (2007) [6] Bolhuis,P.G. et al. Ann. Rev. Phys. Chem. 53, 291-318 (2002) [7] Shan,Y. et al. J. Am. Chem. Soc. 133, 9181-3 (2011) [8] D’Abramo,M., Rabal,O., Oyarzabal,J. and Gervasio,F.L. Angew. Chem. Int. Ed. in press (2012) DOI: 10.1002/ange.201103264

Project Title: LArge scale molecular SImulations of PROtein – DNA recognition in the combinatorial control of trnascription

Project Leader: Dr. Vlad Cojocaru, Heidelberg Institute for Theoretical Studies GmbH, Heidelberg, Germany

Resource Awarded: 2 566 472 core hours on IDRIS – BABEL, IDRIS – Turing, PDC – Lindgren and RZG – SandyBridge


  • Prof. Dr. Hans Schöler – Max Planck Institute for Molecular Biomedicine, Münster, Germany
  • Dr. Matthias Wilmanns – EMBL, Molecular Structural Biology, Hamburg, Germany

In eukaryotic cells, a complex gene expression network is controlled by a relatively small number of protein regulators. How does this limited set of factors regulate the expression of all genes in a complex spatio-temporal pattern is a challenging question. A subset of regulators, the transcription factors bind to regulatory DNA elements and interact either directly or indirectly with other transcription factors or protein regulators. Often, they adapt their conformation to the specific interaction partner and adopt different arrangements on different DNA elements, managing gene expression in a combinatorial fashion. In this project, we aim to reveal the molecular mechanisms of the combinatorial control of gene expression at atomic level of detail. This is essential for understanding a wide variety of biological phenomena. We will focus our study on factors relevant for the maintenance of stem cell ability to differentiate in different cell types and for the reprogramming of somatic cells into stem-like cells (induced pluripotent stem cells). We expect the results will contribute to the elucidation of the mechanisms by which somatic cells can be reprogrammed into pluripotent cells and, more generally, to the understanding of gene regulation. We aim to provide data that may be used in the development of novel stem cell based therapies for a wide range of diseases. To achieve our goals, we will calculate free energy profiles from atomistic molecular dynamics simulations for a set of representative protein-DNA and protein-protein interactions. These calculations are only attainable on large supercomputers due to the large number of degrees of freedom in the simulated system and the extensive sampling of the conformational space required. We will use the highly-parallel NAMD program for the calculations. This application is already available for production runs at several PRACE sites. This project will be performed in close collaboration with experimenters, thus giving us the opportunity to readily test the relevance of our findings.

Project Title: Preferred conformations of paclitaxel at MP2 level

Project Leader: Dr. Marek Łożyński, Poznan University of Technology, Department of Chemical Technology, Poznan, Poland

Resource Awarded: 270 000 core hours on UHEM – UYBHM


  • Maciej Janicki – Adam Mickiewicz University, Department of Biology, Poznan, Poland

Microtubule stabilizing agents (MSA) interfere with disassembly of microtubules in rapidly dividing cancer cells. Paclitaxel (Taxol®), the complex, diterpenoid natural product and the effective MSA, binds to tubulin in a stoichiometric ratio and operates by blocking microtubule dissociation into tubulin dimers finally causes apoptosis i.e. programmed cell death. Although the EC structure of paclitaxel on tubulin Zn+2 sheets showed the location of the ligand, the conformation of paclitaxel molecule was not determined with sufficient precision. The estimation of the bioactive conformation indicated two structures: so called T-taxol, proposed by Emory University, Atlanta, GA group and PTX-NY referred to the New York Sony Brook group. In the present project we address computational questions in order to estimate the conformational space (PES) of paclitaxel as well to judge these two proposals of molecular shape. Another purpose of the present work is to precisely determine the possible hydrogen bond patterns, including cooperative enhancement, and to illuminate the weakly recognized aromatic and olefinic fragments orbital interactions because of such information is important to better understand the nature of non-bonding interactions and their effect on final internal energy. Since density functional methods fail to accurately describe weak π-π interactions, the use of second order Møller-Plesset (MP2) or coupled cluster CCSD(T) (SP) is predicted at cc-pVDZ and aug-cc-pVDZ basis set. The stacking bonding, due to London dispersion energy, will be additionally estimated to characterize attractive interaction energies of benzene-benzene and benzene-ethylene complexes of paclitaxel fragments. We hope that the detailed knowledge on paclitaxel binding site regarding its PES led to the development of second-generation taxanes.

Project Title: MD Simulations of Engineered Light-Switched Chimeric Proteins

Project Leader: Dr Luca Maragliano, Italian Institute of Technology(IIT), , Genova, Italy

Resource Awarded: 3 000 000 core hours on CSCS – Rosa and SURFSARA – Huygens P8


  • Prof Fabio Benfenati – Italian Institute of Technology (IIT), Genova, Italy

Optogenetics is a very recent technique in cell biology which involves the use of genetically encoded, light-gated proteins to perturb and control cellular and organismal behavior in a spatiotemporally precise fashion. A remarkable applications is the modulation of cellular signaling as for example the control of gene transcription and expression. This can be obtained by engineering chimeric proteins where the light-switched domain is properly fused to a protein involved in the transcription process. To realize these chimeras experimentally, atomic detailed knowledge of the proteins structure, relevant interactions, and conformational characteristics is strongly needed. In this project, we will apply and further develop recent computational techniques to efficiently explore the Free Energy landscape describing the conformational properties of engineered light-switched chimeric proteins. Results will provide high-resolution information to guide the experimental realization of the chimeras.

Project Title: Molecular basis of photoreception: new insights from a highly accurate HPC-based modeling

Project Leader: Lars H. Andersen, Anastasia Bochenkova, Aarhus University, Denmark

Resource Awarded: 10 500 000 core hours on HLRS – Laki, LRZ – SuperMig and RZG – SandyBridge


  • Alexander Granovsky – Moscow State University, Russia
  • Kyril Solntsev – Georgia Institute of Technology, Atlanta, USA

The project is aimed at studying photoactive proteins at the atomic level. Photoactive proteins are widespread in nature and enable the signal transduction in biological photoreceptors triggered by the absorption of a photon with a particular wavelength. Opsin proteins containing a protonated Schiff-base retinal chromophore are perhaps the best known as they provide vision in vertebrates. We propose to use the PRACE infrastructure for the computationally demanding modeling of photophysical and photochemical properties of chromophores of these proteins at different levels: isolated chromophore units, then the well-defined atomic-scale interactions within the hosting protein medium, and finally, whole proteins. One of the highlight goals is to understand the catalytic role played by proteins in the ultrafast excited-state reaction dynamics of biological photoreceptors and in the self-regulation of their photo-physical properties. Ultimately, this will lead to an understanding of both wavelength tuning and efficiency of the primary steps in vision.

State-of-the-art electronic structure methods and their large-scale parallel implementations within our original Firefly package, being one of the fastest and the most efficient quantum chemistry packages available today, will be used to track the excited-state evolution of photoactive proteins and their de- excitation pathways through intersection regions of electronic states. Our new extended approach to multistate multiconfiguration quasi-degenerate perturbation theory XMCQDPT2 approved in a number of applications surpasses known methods both methodologically and computationally. The invariance of the new theory makes it unique on the world scale and especially attractive for the purposes of our project. The results of this project will include highly accurate predictions of photoabsorption and emission line shapes of the bare chromophores and photoactive proteins, structures of the minimum energy conical intersections and topographies around them, quantum yields of primary photoinduced reactions and excited-state lifetimes.

Importantly, the project proposed is a joint theoretical and experimental endeavor. As a new initiative on the world-scale, we aim at studying the ultrafast excited-state dynamics in biological systems in vacuum. The fs pump-probe experiments will be carried out to determine, for the first time, the intrinsic response time of the isolated retinal in the protonated Schiff-base form. The project will provide new standards for light-induced protein dynamics and will provide new insight in the molecular basis of vision. It will rely on modern laser-technology and ion-storage techniques which are available in the research group at Aarhus University. Our strategy is to combine expertise of the theoretical team in the field of the state-of-the-art quantum methods with the leadership of Aarhus University when it comes to laser-action spectroscopy techniques for studying the biomolecular ions in the gas phase. We firmly believe that our new joint initiative will open an avenue for the future ground-breaking research in a very broad and highly interdisciplinary field of science, which combines physics, chemistry and biology.

Project Title: Protein Structure Prediction with Biophysical Models

Project Leader: Prof. Wolfgang Wenzel, Karlsruhe Institute of Technology, Institute of Nanotechnology, Karlsruhe, Germany

Resource Awarded: 1 400 000 core hours on CINES – JADE-Harpertown


  • Dr. Ivan Kondov – Karlsruhe Institute of Technology, Steinbuch Centre for Computing, Karlsruhe, Germany

Proteins are the nanoscale machines of all cellular life. Most proteins fold into a unique three-dimensional structure, determined entirely by their amino acid sequence, in a slow and complicated process that is still not fully understood. Because sequencing techniques are far more efficient than structure resolution, predicting the structure of proteins and their functional complexes is one of the major challenges in the life sciences. Closely related is the investigation of protein-protein interactions as one of the most important mechanism of signalling in biological systems.

After the complete genome of many species has been sequenced, there is a huge gap between the number of known sequences and the number of known structures (factor about 1000). Experimental resolution of protein structures is orders of magnitude more expensive than sequencing and not possible for many interesting proteins at all. Computational methods for protein structure prediction are therefore increasingly used to close this gap, but progress for proteins with low sequence similarity to structure the resolved proteins has been slow.

We have developed and biophysics based atomistic simulation techniques for protein folding and protein structure prediction. Using our biophysical all-atom force fields PFF01/PFF02 and efficient massively parallel simulation techniques(Schug, et al., 2006; Schug and Wenzel, 2006; Verma, et al., 2007) we were able to predict the structure of small proteins (up to about 70 amino acids) solely on the basis of the sequence information. Application of this methodology often has direct applications in the life-sciences: We have applied these techniques in close cooperation with experimental groups to elucidate the genetic origin of some human developmental disorders(Kim, et al., 2010; Kim, et al., 2008; Xu, et al., 2011) and also apply them to design inhibitors for protein-protein interactions(Meliciani, et al., 2009).

In this project we will combine these methods with homology modelling techniques and efficient methods for model generation to predict the structure of larger proteins. In a recent CASP competition our group ranked fifth worldwide for template-free structure prediction. POEM is one of the few available biophysical simulation prediction methods and therefore well suited for structure prediction of proteins with low homology to known proteins, but this requires massive computational resources that can only be provided by HPC architectures. In this challenging area there has been little progress in established homology-based structure prediction protocols for the last decade. Progress in the development of these methods will be monitored in the blind comparative assessments of protein structure prediction methods (CASP), where we will participate in 2012 using the resources of this project.

Project Title: Simulating Coupled Protein Folding and Nucleic Acid Binding Using a Polarizable Force Field

Project Leader: Dr Robert Best, University of Cambridge, Department of Chemistry, UK

Resource Awarded: 2 738 400 core hours on PDC – Lindgren


  • Dr Christopher Baker – University of Cambridge, Department of Chemistry, UK

A long-held principle of genomics is that a protein’s amino acid sequence determines its structure, and that its structure determines its function.1 While this may be true for a large number of proteins, it is now known that there are a significant number of proteins that are both intrinsically disordered and of biological and medical significance. These intrinsically disordered proteins (IDPs) are natively unfolded, or possess long, unstructured regions.2 In addition to having no structure, it seems that many IDPs actually rely on their lack of structure for their function and are able to bind specifically and selectively to partners including proteins and nucleic acids.3,4,5,6,7 In fact, it has been speculated that a lack of structure facilitates binding. According to the “fly casting” mechanism,8 a protein has a much larger capture radius in its unfolded state than in its folded state and so is able to make initial contact with its binding partner at a larger separation, subsequently folding as it is drawn in towards its partner. It is also possible that IDPs are able to adopt different folds to recognize different binding partners. While there is now some experimental evidence to support the “fly casting” hypothesis,9 determining the mechanism by which coupled protein folding and binding occurs at the atomic level remains a challenge.10 The fundamental question that we will address in this project is: What is the molecular mechanism of coupled protein folding and nucleic acid binding? With such details difficult to access directly by experiment, computer simulation is a powerful tool that allows us to sample coupled folding and binding events with atomic resolution. The atomistic simulation of coupled folding and binding, however, is at the limit of what is possible using current molecular dynamics (MD) simulation technologies, and has not yet been achieved. Moreover, the binding of proteins to nucleic acids is a process dominated by electrostatic interactions, yet the majority of force fields currently available for MD simulation do not provide an adequate representation of electrostatics – particularly changes induced by the approach of highly charged proteins and nucleic acids. We intend to use a state of the art, polarizable force field11 to address the question above with the specific aim of: • using all-atom molecular dynamics simulations to determine the atomic-resolution mechanism of coupled protein folding and nucleic acid binding. Specifically, we will simulate coupled folding and binding between the Jun-Fos heterodimer and DNA.6 Jun and Fos are components of the transcription factor AP1, which regulates gene expression in response to stimuli. They are expressed in a range of tissues and recognize, as either the Jun-Jun homodimer or Jun-Fos heterodimer, DNA sequences that occur frequently in the human genome. As potent activators of mitogenic activity, the hyperactivity of Jun and Fos shows a positive correlation with oncogenic cell transformations: as well as providing general insights into the binding of IDPs to DNA, an improved understanding of how interactions between Jun-Fos and DNA could be inhibited may open the door to novel anti-cancer therapies.

Project Title: Simulating the active response of cells to external mechanical stimuli

Project Leader: Dr. Patrick McGarry, National University of Ireland, Galway, Galway, Ireland

Resource Awarded: 945 000 core hours on ICHEC – Stokes

Numerous studies have shown that stress fibres in a cell will align in a dominant direction when subjected to different mechanical loading conditions. Cells subjected to fluid shear will form stress fibres aligned along the flow direction. Cells adhered to a substrate which is subjected to cyclic stretch will form stress fibres aligned perpendicular to the direction of stretch. Furthermore, cells on a cyclically stretched substrate will realign if subjected to uniaxial stretching and the extent that the stress fibres align in a dominant direction is dependent on the stretch magnitude. In contrast, cells in the 3D environment of a collagen gel will align parallel to the direction of applied cyclic stretching. The mechanisms underlying this behaviour are poorly understood and previous computational investigations have considered purely elastic fibres or stochastic models that do not consider the underlying biochemistry. Such approaches offer limited insight into the cellular mechanisms that drive stress fibre and cell alignment. These phenomena have important implications for atherosclerosis and post angioplastly restenosis. The cyclic behaviour of cells also has implications for medical device design, e.g. the effect of stenting on the response of cells to pulsatile loading, and for the tissue engineering in a dynamic loading environment. As part of the on-going research sponsored by Science Foundation Ireland an advanced computational framework that will capture the response of cells to dynamic loading being developed by considering two mutually dependent cellular processes; i.e. the formation of contractile stress fibres in response to mechanical and chemical stimuli and the formation of traction dependent focal adhesions. Model development will include 3D implementation of a constitutive formulation that predicts the distribution and contractility of stress fibres. This implementation will be used to predict the behaviour of cells subjected to static loading such as micropipette aspiration and parallel plate compression. A thermodynamically motivated model for focal adhesion formation will be used in tandem with this stress fibre formulation to predict the static response of cells spread on different elastic substrates. Further development of these formulations will also necessary to capture the behaviour of dynamically loaded cells, including the development of a fading memory constitutive formulation for stress fibre contractility.

Project Title: The Molecular Basis for Ligand Recognition & Signalling in Toll-Like Receptors via Simulation

Project Leader: Dr Peter Bond, University of Cambridge, Department of Chemistry, UK

Resource Awarded: 3 200 000 core hours on CSCS – Rosa

A century ago, Elie Metchnikoff received the Nobel Prize in Medicine and Physiology for recognizing the innate ability of organisms to detect pathogens through a cellbased defensive system. But the “sensors” of the associated pathogenic molecules, the Toll-like receptor (TLR) proteins, were only identified within the last decade; indeed, this year’s Nobel Prize has just been awarded to Jules Hoffmann and Bruce Beutler for their discovery of TLRs and their role in recognition of pathogenic signals. TLRs represent the initial gateway to almost all mammalian inflammatory responses to invading microbes. They are specialized for binding ligands with diverse structural and physiochemical properties, ranging from microbial cell wall components to nucleic acids, and there is substantial interest in their pharmacological manipulation. Of particular biomedical interest is TLR4, which binds bacterial outer-membrane lipopolysaccharide (LPS) with nanomolar sensitivity. Minute amounts of LPS released from invading bacteria are an early sign of infection and prepare the immune system to counteract further illness. Recognition of LPS by the TLR4 complex leads to signalling within the cell interior to initiate an immunological response. However, if this response is uncontrolled, it can cause septic shock, the number-one cause of intensive care unit deaths worldwide. Furthermore, the association of TLR4 with many infectious, allergic, inflammatory, and malignant diseases emphasizes its importance as a therapeutic target. There is great variation between bacterial species in the structures of naturally occurring TLR4 ligands, yet even subtle structural alterations can profoundly affect activity, leading to unpredictable changes in the immunological response. This makes the design of safe and effective therapeutic drugs targeted to the TLR4 pathway problematic. High-resolution structural information is available for the activated receptor complex, providing some clues to the basis for of ligand recognition, but yielding only limited information on the detailed mechanisms of receptor regulation. Traditional structure-based drug design approaches are hampered by the large, complex, and flexible nature of LPS analogues. Therefore, in this project we will use a simulation approach to probe the molecular determinants of TLR4 ligand recognition and activation, towards the development of novel therapeutic treatments. Explicitly-solvated, all-atom molecular dynamics simulations will be performed for the entire, signalling-competent receptor complex in the presence of a range of bound ligands, with typical systems amounting to approximately half a million atoms. Long-timescale trajectories will provide a means to systematically probe the dynamic effects of a ligand library on receptor stability. Furthermore, state-of-the-art free-energy calculations will yield an understanding of the energetic determinants of recognition and binding. Finally, multiscale simulation approaches will be used to predict the higher order structure of ligands, which is likely to affect bioavailability within the body. At the same time, our data will be used to iteratively guide live cell screening and spectroscopic experiments. The result will be a systematic rationalization of the molecular mechanisms of TLR4 recognition and signalling, towards the rational design of novel immunomodulatory therapeutic treatments for a host of infectious and inflammatory diseases.

Project Title: Virtual Imaging Platform of virtual physiological Human

Project Leader: Prof. Denis Friboulet, INSA Lyon, CREATIS, Lyon, France

Resource Awarded: 1 670 670 core hours on EPCC – HeCToR XE11


  • Dr. Tristan Glatard – INSA Lyon, CREATIS, Lyon, France
  • Mr. William Alberto Romero Ramirez – INSA Lyon, CREATIS, Lyon, France
  • Mr. Stefan Zazada – UCL, London´s Global University, UK

The Virtual Imaging Platform (VIP) is a project of the French National Research Agency providing an online, open platform to simulate medical images in 4 modalities, namely Magnetic Resonance Imaging (MRI), ultrasound imaging, Positron Emission Tomography (PET), and Computed Tomography (CT). More information about the project is available at The platform is currently connected to the European Grid Infrastructure to support embarrassingly parallel simulations, but it lacks reliable resources to run parallel codes, in particular MPI. Therefore, the motivation of project `Virtual Imaging Platform for Virtual Physiological Human` (VIP for VPH) is to offer imaging scientists a convenient mechanism to perform image simulations on PRACE High-Performance Computing (HPC) resources. This will be demonstrated on a 3D+t 512x512x512 simulation of a Magnetic Resonance Imaging acquisition (MRI), and made available for VIP users. The VIP workflow engine is being interfaced with the Application Hosting Environment (AHE) that can launch jobs on PRACE resources. AHE will be in charge of data transfers, job submission and monitoring to PRACE.

Earth Sciences Projects

Project Title: Algorithmic parameter estimation of large-scale prediction models

Project Leader: Heikki Järvinen, Finnish Meteorological Institute, Helsinki, Finland

Resource Awarded: 4 200 000 core hours on EPCC – HeCToR XE9


  • Heikki Haario – Lappeenranta University of Technology, Finland
  • Erkki Oja – Aalto University, Espoo, Finland

Modern society is profoundly reliant on numerical simulation models in areas such as weather forecasting and climate research. More powerful computers and better understanding of the Earth science has led to increasingly complex simulation tools. It is important that the uncertainties of the simulation results are well understood, because the information is used in decision making in the society.

Practically all discrete numerical simulation models contain closure schemes where some unresolved, or sub-grid scale, quantities are expressed using specified parameters rather than some explicit modelling. The PARAMETER project aims at applying advanced parameter estimation techniques (i) to quantify the modelling uncertainty related to these model parameters, and (ii) to tune the predictive skill of the models by means of algorithmic model parameter estimation. Specifically, the PARAMETER project will focus on tuning the predictive skill of numerical weather prediction models, based on the methods published recently by the research team.

The scientific research behind the methods to be applied in the DECI project PARAMETER is funded by the NOVAC project of the Computational Science Research Programme of the Academy of Finland, the Nessling foundation, the EU/FP7 project EMBRACE (, and the Academy of Finland Centre of Excellence in Inverse Problems.

Project Title: Regional Climatic Research using a Representative Carbon Pathways Approach

Project Leader: Dr. Rodney Teck, National University of Ireland, Maynooth, Maynooth, Ireland

Resource Awarded: 842 400 core hours on PDC – Lindgren


  • Dr. Adam Ralph – ICHEC, Dublin, Ireland

A key limitation of Global Climate Models (GCM’s) is the fairly coarse horizontal resolution, typically 3.75o by 2.5o (500 km x 300 km). For practical planning, countries require information on a much smaller scale than GCM’s are able to provide. One of the solutions is to embed a Regional Climate Model (RCM) in the GCM. The aim of this project will be to use the output of the most up to date European Global Climate Model, EC-Earth and dynamically downscale the data through the use of a Regional Climate Model to a scale more useful to local planning (i.e. water resource and flood defences). In the past GCM’s have relied upon hypothetical scenarios, Special Report on Emission Scenarios (SRES) based on four future predictions of world development: A1, A2, B1, B2. The SRES scenarios, however, do not encompass the full range of possible futures: emissions may change less than the scenarios imply, or they could change more. Current thinking within the Intergovernmental Panel on Climate Change (IPCC) has resulted in a new approach to scenarios: choosing a handful of emission trajectories, known as Representative Carbon Pathways (RCP’s). The RCP’s then became the basis for a series of new climate runs in the latest climate models, such as the EC-Earth GCM. The new choice of scenarios has four emission trajectories to focus on and have labelled them based on how much heating they produce at the end of the 21st century – 8.5, 6, 4.5 and 2.6 watts per square metre (Wm-2). At the high end, the 8.5 Wm-2 case, carbon dioxide levels soar to 1,300 parts per million by the end of the 21st century – and are set to rise even further. The project’s aim will be to dynamically downscale the RCP’s 6 and 8.5 Wm-2 output from the EC-Earth GCM for the periods: 1961 – 1990 – this will be used as an historical reference base and the 2005 – 2021 data for future predictions. This will be carried out using The National Center for Atmospheric Research (NCAR’s) Weather Research and Forecasting Model (WRF) used as a regional climate model (RCM).

Engineering Projects

Project Title: Future-proof High Performance Numerical Simulation for CFD with FEATFLOW

Project Leader: Dipl.-Inf. Markus Geveler, TU Dortmund University of Technology, Institute for Applied Mathematics, Dortmund, Germany

Resource Awarded: 5 990 400 core hours on CINECA – PLX


  • JProf. Dominik Goeddeke – TU Dortmund University of Technology, Institute for Applied Mathematics, Dortmund, Germany
  • Prof. Stefan Turek – TU Dortmund University of Technology, Institute for Applied Mathematics, Dortmund, Germany

We aim at the development of efficient, reliable and future-proof numerical schemes and software for the parallel solution of partial differential equations (PDEs) arising in industrial and scientific applications. Here, we are especially interested in technical flows including Fluid-Structure interaction, chemical reaction and multiphase flow behaviour which can be found in a wide variety of (multi-) physics problems. In our approach, both numerical and hardware efficiency are addressed simultaneously: On the one hand, the transition of today’s computational hardware towards parallel (and heterogeneous) architectures is in progress and therefore all levels of parallelism (vectorisation, parallelism on the core level in multi- and many-core CPUs and accelerator devices like GPUs and finally on the node level within distributed memory clusters) have to be exploited. Algorithms and whole solvers have to be tailored with respect to the target hardware in order to achieve a significant amount of the parallel peak performance. On the other hand, the sole concentration on hardware efficiency does not carry out the whole job (and in some cases may be counter-productive): Numerical efficiency plays a crucial role and itself includes multiple levels that can be optimised. Starting with the overall numerical and algorithmic approach required for the solution of a given domain specific problem (i.e. discretisation of the governing equations in time and space), stabilisation, linearisation of non-linear problems and finally the solution of the linear problems and smoothing therein, all these aspects together with the aforementioned levels of parallelism bear a large amount of interdependencies. In the proposed project, we want to improve our parallel numerical software framework FEATFLOW ( and augment this powerful simulation toolkit for academic and industrial usage. The planned improvements take both aspects of efficiency into account in order to make it ready for future HPC-architectures: Novel numerical- and physics-components as well as software-techniques for massively parallel (heterogeneous) compute resources are going to be employed that extend the applicability of the package to current and future real-world problems in the field of CFD.

Project Title: Lattice Boltzmann based simulations of flowing soft systems

Project Leader: Prof Massimo Bernaschi, National Research Council(CNR), Instituto per le Applicazioni del Calcolo, Italy

Resource Awarded: 1 597 440 core hours on CINECA – PLX


  • Prof Roberto Benzi – University of Rome Tor Vergata, Dipartimento di Fisica, Italy
  • Dr Mauro Sbragaglia – University of Rome Tor Vergata, Italy
  • Dr. Sauro Succi – IAC-CNR, Roma, Italy

The rheology of flowing soft systems, such as emulsions, foams, gels, colloidal glasses and related complex fluids, is a subject of increasing importance in modern non- equilibrium thermodynamics, with a broad range of applications in fluid dynamics, chemistry and biology. From the theoretical standpoint, flowing soft systems are challenging because they do not fall within any of three conventional states of matter, gas-liquid-solid, but live rather on a moving border among them. Foams are typically a mixture of gas and liquids, whose properties can change dramatically with the changing proportion of the two; wet-foams can flow almost like a liquid, whereas dry-foams may conform to regular patterns, exhibiting a solid-like behavior. Emulsions can be paralleled to bi-liquid foams, with the minority species dispersed in the dominant (continuous) one. The behavior and, to same extent, the existence itself of both foams and emulsions are vitally dependent on surface tension, namely the interactions that control the physics at the interface between different phases/components. Living, as they do, out of equilibrium, these materials exhibit a number of distinctive features, such as long-time relaxation, anomalous viscosity, aging behavior, which necessitate profound extensions of non-equilibrium thermodynamics. The study of these phenomena sets a pressing challenge for computer simulations as well, since characteristic time-lengths of disordered fluids can escalate tens of decades over the molecular time scales. Among a variety of numerical methods for complex flows, both atomistic and macroscopic, mesoscopic lattice Boltzmann (LB) models have recently been developed, which prove capable of reproducing a number of qualitative features of soft-flowing materials, such as slow relaxation, dynamical heterogeneities, aging and others. These models are based on suitable generalizations of the multicomponent Shan-Chen scheme for non-ideal fluids, with multi-range competing interactions, namely short-range attractions plus mid-range repulsion. The competition between short-range attraction and mid-range repulsion lies at the heart of the very rich behavior of the density field. Owing to this complexity, and particularly the slow relaxation properties, the investigation of the dynamical behavior of these systems requires very long time integrations, typically of the order of tens of million of time steps (as a reference, one LB time step can be taken of the order of 100 1000 molecular dynamics time-steps). As a result, the need of a very-long simulation span, sets a strong incentive for efficient implementations. To that purpose we developed a multi-GPU CUDA implementation of this extended LB model that provides very significant time gains with respect to our previous, albeit highly-tuned, CPU implementation. The expected outcome of the numerical experiments we would like to run within the Prace initiative is a better understanding of the complex flow properties of foams and real emulsions, from mesoscale up to their hydrodynamical behavior and rheological properties. In particular, the idea is to improve LB models to describe emulsions and their properties under confinement in different geometries at the changing of the boundary conditions (chemical patterning or geometrical roughness).

Project Title: A Numerical Analysis of Hydrogen Underexpanded Jets

Project Leader: Prof Vinicio Magi, University of Basilicata, Department of Environmental Engineering and Physics , Potenza, Italy

Resource Awarded: 1 720 320 core hours on CINECA – PLX and WCSS – WCSS Supernova


  • Dr Francesco Bonelli – University of Basilicata, Department of Environmental Engineering and Physics , Potenza, Italy
  • Prof Annarita Viggiano – University of Basilicata, Department of Environmental Engineering and Physics , Potenza, Italy

The aim of this work is the study of the fluid dynamic behavior of underexpanded hydrogen jets by using a High Performance Computing (HPC) methodology. Sonic and supersonic gas jets are of interest from the point of view of both fundamental fluid mechanics and engineering applications, such as engines, combustors, etc. A comprehensive understanding of the global structure and of the dynamics of these jets is important both for the development of new strategies either to improve the performance and efficiency of direct-injection engines or to verify the feasibility of new propulsion systems and for safety issues. In this scenario, this work is aimed to study the fluid dynamic behavior of underexpanded hydrogen jets for ground applications. Hydrogen is the most attractive fuel for Internal Combustion Engines (ICEs) since hydrogen-fueled ICEs can work with near-zero emissions and higher efficiencies than conventional hydrocarbon-fueled ICEs [1]. These capabilities are due to the unique features of hydrogen compared to conventional fossil fuels. The wide flammability limits allow stable combustion with very lean mixtures, thus resulting in a reduction in maximum temperatures and, consequently, in nitrogen oxides (NOx). Moreover, the engine can operate unthrottled at low-loads with an improvement of engine efficiency. On the other hand, hydrogen is responsible for some inconvenience at high engine loads. Specifically, the low ignition energy can cause undesirable combustion events, such as preignition, knock and backfire, whereas the small quenching distance leads to narrow thermal boundary layer, with higher thermal losses. Finally, the low density causes a reduction of volumetric efficiencies. Certainly, one of the most promising strategy in order to overcome these problems is the Direct Injection (DI) strategy. This approach requires high pressure and high-flow rate injectors to ensure a good mixing in the short time available, therefore, hydrogen injection is designed to be sonic with pressure higher than 80 bar [1]. This results in an underexpanded jet issuing into the cylinder. The structure of underexpanded jets is well documented [2,3] and its analysis has been addressed by several authors [4–7] by using both experimental measurements and numerical simulations. In this work the analysis will be carried out by employing a two-dimensional axial symmetric in-house code [8] which is able to take into account real gas effects by employing either Van der Waals or Redlich-Kwong equations of state (EoS). The accuracy of the computational model has been assessed by comparing the results with those obtained experimentally by Wang and Andreopoulos [9] and by Woodmansee et al. [5]. The first aim of this project is to evaluate the influence of real gas effects on hydrogen jets structure. Infact, very recent works [10,11] have shown the importance to employ real gas equations dealing with highly underexpanded hydrogen jets. Therefore, in order to select the most suitable EoS the numerical results will be compared with those obtained experimentally by Tsujimura et al. [12]. Finally a parametric analysis will be carried out by varying the injection pressure and temperature, in order to investigate the features of underexpanded jets and real gas effects under different conditions.

Project Title: Large scale simulation of turbulent pipe flow

Project Leader: Dr. Philipp Schlatter, KTH, Department of Mechanics, Stockholm, Sweden

Resource Awarded: 6 250 000 core hours on CSC – Louhi XT and EPCC – HeCToR XE10


  • Dr. Geert Brethouwer – KTH, Department of Mechanics, Stockholm, Sweden
  • Prof. Elisabetta De Angelis – Università di Bologna, CI RI Aeronautica, Bologna, Italy
  • Dr. George El Khoury – KTH, Department of Mechanics, Stockholm, Sweden
  • Dr. Paul Fischer – Argonne National Laboratories, Mathematics and Computer Science, Argonne, USA
  • Prof. Arne V. Johansson – KTH, Department of Mechanics, Stockholm, Sweden
  • Prof. Alessandro Talamelli – Università di Bologna, CI RI Aeronautica, Bologna, Italy

The flow of fluids in pipes with circular cross-sections is frequently encountered in a variety of environmental, technical and even biological applications. Typical examples of pipe flows can be found in urban drainage systems, transport of natural gas or oil in the energy sector, or the flow of blood in veins and arteries. Accordingly, the understanding of flow physics in pipes has a direct and substantial impact on everyday life and an adequate knowledge of such flow problem will help in finding scientific methods to reduce drag and the like. The Navier-Stokes equations govern the dynamics of turbulent flows. This set of equations, when properly non-dimensionalized, includes the Reynolds numbers (Re) as the main characterization. Re is by far the most important non-dimensional number in fluid mechanics, and can be considered as a measure for the “speed” of the flow inside the pipe. Turbulence is a characteristic state of flows with sufficiently high speeds, or high Reynolds numbers. Most fluid flows observed in nature are indeed turbulent. Of particular importance in flows delimited by solid walls is the near-wall region in which a large fraction of the drag stems from velocity fluctuations in a thin boundary layer adjacent to surfaces. Near-wall turbulence structures in wall-bounded shear flows primarily scale in terms of the so-scaled viscous length scale, which might be very small as Re is increased. However, according to recent experimental studies, very large-scale motions with lengths of 5R up to 20R are found in fully developed turbulent pipe flow (R being the radius). These structures, being strongest in the outer region, even extend throughout the layer and even leave their footprint quite close to the wall. These large-scale structures are very energetic and active. Large-scale motions thus play an important role in the dynamics of turbulent pipe flows. The aim is to study fully developed high-Reynolds number turbulent pipe flow through direct numerical simulations (DNS). DNS attempts to resolve all relevant scales of the turbulent flow. These will be carried out using the massively parallel DNS code available at KTH Mechanics, nek5000, which is based on an accurate and efficient spectral-element discretization. Pipe flow is the case which is easiest realisable in experiments. However, due to numerical difficulties related to the cylindrical coordinates and the corresponding numerical singularity arising along the symmetry line, it is the only canonical flow case that has not yet been thoroughly studied using DNS, as opposed to plane channels and boundary layers. The proposed work will be a complement to both the ongoing simulations of high-Reynolds number turbulent boundary layers and channels, but also relevant in conjunction with the forthdoming experimental studies of high-Reynolds number turbulent pipe flow within the CICLoPE project (CICLoPE consists of a 120 m long pipe with diameter 0.9m, situated near Bologna/Italy). To obtain a database for turbulent pipe flow that is well validated with both experiments and other simulations is very timely. Therefore, we intend to use this present DECI project to complement our simulations at lower Reynolds numbers with dataset pertaining to higher Re. Specifically, we would like to be able to present data in the range of Re=180 to 1000, all obtained in a long pipe employing the same high resolution. Such data will be valuable not only to increase our knowledge of generic wall turbulence, but will also allow for the development of better models of turbulence for industrial applications.

Project Title: Fluid-structure interaction for Smart Wing Design

Project Leader: Dr. Marianna Braza, Institut de Mécanique des Fluides de Toulouse, France

Resource Awarded: 5 600 000 core hours on ICHEC – Stokes and WCSS – WCSS Supernova


  • Dr. Yannick Hoarau – Université de Strasbourg, Institut de mécanique des fluides et des solides de Strasbourg, Strasbourg, France

This project concerns the analysis by numerical simulations of compressible flows or flows with variable density around aerofoils in 2D or 3D. We aim to better know the physical mechanisms related to the arising instabilities and the unsteadiness and the transition in flows around obstacles since the very first steps towards turbulence at moderate Reynolds numbers in order to better model these mechanisms at high Reynolds number.

Project Title: Inertial particles in transitional flows

Project Leader: Dr. H.C. de Lange, Technical University of Eindhoven, The Netherlands

Resource Awarded: 4 200 000 core hours on PDC – Lindgren


  • Dr. L. Brandt – KTH, Sweden
  • PhD. Joy Klinkenberg – Technical University of Eindhoven, The Netherlands
  • Dr. Gaetano Sardina – Universita KORE di ENNA, Facolta’ di Ingegneria, Arhcitettura e delle Scienze motorie, Enna bassa, Italy
  • Dr. Philipp Schlatter – KTH, Department of Mechanics, Stockholm, Sweden
  • Prof. Anton van Steenbergen – Technical University of Eindhoven, The Netherlands

Traditionally, direct numerical simulations (DNS) of transitional and turbulent flows are performed in simplified computational domains that are characterised by periodic boundary conditions in all three directions. However, real applications in nature and technology often involve the interaction with a solid wall and are thus inhomogeneous is space. Here, we study the flow case of a transitional spatially evolving boundary layer exposed to free-stream turbulence, as typically observed on turbine blades. We focus on the advection of small inertial particles in transitional intermittent flows. Such a computational study based on highly resolved DNS, not yet attempted in the literature, bears many interesting and relevant physical effects due to the growing boundary layer and to the random appearance of regions of laminar and turbulent flows; for instance the non-dimensional number characterising the particle-wall accumulation is gradually changing with the downstream distance. The raw scientific data will be shared with the scientific community (iCFDdatabase,

From a computational point of view, transitional flows are necessarily investigated in very long domains in order to accurately capture the whole transition process; in addition, the intermittent nature of the flow requires long statistical sampling. It is thus only recently that the DNS of transitional boundary layers has become feasible. The additional complexity of coupling the advection of inertial particles leads to large computational demands on massively parallel computers.

Materials Science Projects

Project Title: Comprehensive Ab initio studies of Nitride and Oxide fuels and Nuclear Structural materials

Project Leader: Dr. Pär Olsson, KTH, Department of Physics, Stockholm, Sweden

Resource Awarded: 3 125 000 core hours on CSCS – Rosa


  • PhD Student Zhongwen Chang – KTH, Department of Physics, Stockholm, Sweden
  • PhD Student Antoine Claisse – KTH, Department of Physics, Stockholm, Sweden
  • PhD Student Luca Messina – KTH, Department of Physics, Stockholm, Sweden
  • PhD Student Merja Pukari – KTH, Department of Physics, Stockholm, Sweden
  • PhD Student Odd Runevall – KTH, Department of Physics, Stockholm, Sweden

For future generation nuclear power plants, fission or fusion based, the need for improved fuels and structural materials is crucial. The use of nuclear power is associated with several major problems: the handling of the long-lived radioactive waste, the limited resources of U-235 and the safety and integrity of the structural materials. These issues are addressed by the development of advanced reactor types, GenIV reactors, in which Am and Pu are transmuted, thereby decreasing the effective half-life of the waste, and at the same time fissile fuel is generated from natural uranium (U-238). This is achieved by the use of non-moderated (fast) neutrons in the fission process, and so called fast reactors have been used on an experimental scale for decades. A fundamental challenge connected to the use of fast neutrons is the damage induced in various parts of the reactor. Well known effects are hardening, embrittlement and swelling of the fuel and of internal parts such as cladding, spacers, tubes, and of the reactor vessel. Much effort has been put into experimental studies and modeling of radiation damage since its discovery about 50 years ago. The atomistic mechanisms behind hardening and swelling in metals and alloys are today well understood, although predictive modeling is difficult to achieve because of the complexity of these processes. In contrast, the fundamental mechanisms of radiation damage in typical fuel matrices, like uranium-oxide or, more importantly, in innovative fuels like metal-nitride matrices, are not well understood. In the current project, we will use first-principles electronic structure calculations to study the structural and thermal properties of radiation induced defects in ceramic fuels, and their interaction with transmutation gas atoms line He and Xe. We will also use first principles molecular dynamics to study the fundamental mechanisms of defect formation in ceramic crystals. Clearly, the investigation of how intense neutron radiation affects these fuels must be based on experiments. However, theoretical modeling, e.g., atomistic simulations may provide a firmer footing for the interpretation of such experiments, and in that sense plays, and will play, a very important role in the development of future advanced fuels. We are involved in two EU projects, in which experiments and modeling are essential parts. For the modelling of the structural materials, special focus will be devoted to modelling the behaviour under irradiation of Oxide Dispersion Strengthened (ODS) steels. This new class of nano-structured materials has shown good resistance to radiation effects. However, these novel materials have not, and cannot, be tested in real life for timescales corresponding to the lifetime of a reactor. Therefore, we here propose to model the ageing and degradation of the mechanical properties of the ODS steels and other nano-structured materials. The modelling will be based on existing experimental data and on first principles quantum mechanical calculations, feeding these data into higher scale models, such as kinetic Monte-Carlo, where the long term evolution of the alloy microstructure can be simulated. The most compute intensive part of this project consists of the first principles calculations. The ODS and nano-structured materials have been shown in many studies to have a very good response to neutron irradiation. Especially the irradiation induced swelling and creep are minimal. However, these studies have all been performed for short time spans where the nano-clusters have no time to disintegrate and diffuse into the matrix or to grain boundaries or other sinks. Therefore, it is of critical importance to evaluate what will happen over longer times. The stability, under irradiation, of these nano-clusters can be assessed using a multi-scale modelling approach. The largest computational load of these two efforts will be to perform first principles calculations using the VASP code, which has been optimized on numerous architectures, including Cray XE6.

Project Title: Collective Motions in Protein Crystals

Project Leader: Prof. lyndon Emsley, Ecole Normale Supérieure de Lyon, France

Resource Awarded: 330 000 core hours on RZG – Genius


  • Dr. Maria Baias – Ecole Normale Supérieure de Lyon, France
  • Dr. Martin Blackledge – CEA, France
  • Prof. Stephan Grzesiek – University of Basel, BIOZENTRUM, Basel, Switzerland
  • Dr. Jozef Lewandowski – University of Warwick, UK

Protein motions occurring in the nano- to millisecond time scale are of great relevance to protein function, and a good understanding of protein dynamics is crucial for understanding numerous biophysical processes occurring in these complex systems, such as enzymatic catalysis, molecular recognition, ligand binding, and protein folding. To probe this, we use solid-state NMR spectroscopy, which allows access to site-specific information about biomolecular motions. Different NMR experiments are sensitive to motions from different timescales, allowing us to investigate fast protein dynamics of the order of few nanoseconds to slower dynamics taking place in the miliseconds time scale. Many approaches exist to model dynamics in proteins to explain the NMR experimental data, and in order to validate any of these models a new method must be implemented to confirm the existing results. In particular, the milisecond regime is very challenging to investigate, and methods are not completely established for analyzing and interpreting NMR data recorded for slow motions in proteins. The most unbiased method would be provided by computational approaches, such as performing molecular dynamics calculations on a time scale that would allow us to probe slow motions in proteins. We plan to develop this method, and we will consider two model protein systems, the relatively small immunoglobulinbinding protein G (GB1) [1-3], and the more complex dimeric 153-residue microcrystalline ZnII-loaded human superoxide dismutase (ZnII-SOD) [4]. For these systems, extensive experimental NMR data have recently been acquired in microcrystalline form, and tentatively interpreted in terms of slow local motions. Performing the MD calculations would allow us to confirm this interpretation, and provide a deeper insight into these slow dynamic processes.

Project Title: First-Principles Calculations on the Conformation Transfer in a Dual-core Molecular Switch

Project Leader: Prof. Mats Persson, University of Liverpool, Liverpool, UK

Resource Awarded: 3 040 000 core hours on PDC – Lindgren


  • Dr Felix Hanke – University of Liverpool, Liverpool, UK

The design and realization of switches on the atomic and molecular scale is one of the grand challenges in nanotechnology research. For instance, the ability to store and retrieve bits in single molecules, would lead to a computer memory at the ultimate space limit. The main challenges in the field of molecular switches are related to the writing, addressing, and information transfer at the atomic and molecular scale. These challenges are addressed in our EU future and emerging technology (FET) open network ARTIST, which involves some of the most prominent experimental groups in nanoscience. The CONTRAR project addresses some of these challenges with large scale computer simulations of a dual-core molecular switches. Here we are investigating a redox- and conformation based molecular switch, e.g. where the addition or subtraction of an electron leads to significant changes in the molecule itself, which makes it very easy to read out the on- or off-state. In this particular study, we will go beyond the study of molecules with a single charge centre, by investigating a recently synthesized molecule with two charge centres that s capable of storing up to two electrons. This highly challenging and ambitious computational project is at the very core of our ARTIST network. We want to investigate the mechanisms by which the different charge and conformational states are stabilized, which in return can provide an avenue to localize and control the transfer of single electrons at a scale of around 1 nm. The ability to control the motion of single electrons within a single molecule is one of the key challenges that needs to be overcome in the quest of single-molecule electronics. It is expected that the results from this study can be generalized to redox switches with more than two separate charge centres. We have already undertaken preliminary studies on single redox-switches of the same type and we have already shown that an accurate simulation of these molecules needs to account for the van der Waals forces behind the adsorption process, and provide a proper description of the electron correlation at the metal centre. Th is very difficult to achieve with current density functional methods and can only be done with a van der Waals-density functional combined with a GGA+U approach. On top of all that, it is vital to model almost the entire experimental setup, including the molecule itself, the ultrathin supporting insulator, as well as the metal underneath which forms a charge-reservoir and which is required to fix the electronic chemical potential of the entire system.

Project Title: Electronic structure and orbital effects in polynuclear clusters

Project Leader: Prof. Dr. Grzegorz Kamieniarz, University Poznan, Department of Physics, Poznan, Poland

Resource Awarded: 810 000 core hours on SURFSARA – Huygens P6


  • Associate Professor Leeor Kronik – Weizmann Institute of Science, Department of Materials and Interfaces, Rehovot, Israel

Polynuclear magnetic clusters and chains behave like individual quantum nanomagnets, displaying quantum phenomena on macroscopic scale. The accurate simulation of these complex objects becomes the key issue not only in order to understand their fascinating properties but also because of their potential applications in magnetic storage devices or in envisaged quantum computer processor as well as in the low-temperature refrigerants. The magneto-structural correlations, the role and mechanism of magnetic anisotropy and intrinsic quantum effects following from the geometrical frustration induced by the topological arrangement of spins or particular interactions count among the new challenges for computer simulations. The chromium-based rings and dimers are outstanding materials for quantum information processing and for low-temperature cooling. Although their properties are cleared up to some extent on the phenomenological level, they still need more fundamental understanding on the basis of electronic structure calculations with the orbital degree of freedom taken into account. The real challenges appear III for the molecules grafted on a surface or containing more than eight Cr S=3/2 ions and/or are doped II II by magnetic Ni or Cu ions. As soon as both the magnetic and nonmagnetic impurities or bond defects appear, the number of relevant spin configurations increases substantially and implies an enhanced demand on the computing resources. We expect that within the project we explain and calculate the magnetic interactions between the chromium ions and other 3d ions in agreement with experiment and justify the stability of the chromium-based dimers recently synthesized as the candidates for a pair of quantum bits as well as the stability of the chromium rings on the gold surface. The efforts to stabilize the polynuclear clusters on surface is a challenge in the field of addressing molecular nanomagnets as envisaged storage memory units. The substrate may become devastating for the geometrical structure and magnetic properties which are correlated.

Project Title: Multifunctional biomarkers for electron paramagnetic resonance imaging

Project Leader: Dr. Zilvinas Rinkevicius, KTH, School of Biotechnology, Stockholm, Sweden

Resource Awarded: 1 875 000 core hours on CSCS – Rosa

Alzheimerʼs disease is one of the most prominent cause of the acquired dementia in elderly patients and it affects around 35.6 million people worldwide. In Sweden among the 160 thousands with dementia around 45% have been diagnosed with Alzheimerʼs disease. The Alzheimerʼs disease have the profound impact on the patients and their families and the overall impact of this disease on the whole society is expected to increase in the future with the population aging in Europe. Early diagnostics of the Alzheimerʼs disease is essential for efficient treatment of this disease and efficient screening of the people within risk groups. Unfortunately, currently options for clinical diagnostics of early stages of the Alzheimerʼs disease is very limited and development of novel clinical imaging techniques are highly desirable. Present research project aims to address this problem and focuses on the development of the electron paramagnetic resonance imaging technique, which is promising methodology for in vivo imaging of early damage to brain tissue cause by Alzheimerʼs disease. Within this project we aim to develop novel fluorescent spin labels, which are employed as the contrast agents in the electron paramagnetic resonance imaging, using the state of the art molecular modeling tools.

Project Title: Multiscale Modelling of Organic LED

Project Leader: Dr. David Beljonne, University of Mons, Chemistry of Novel materials, Mons, Belgium

Resource Awarded: 1 708 700 core hours on CSC – Sisu and ICHEC – Stokes


  • Dr. Thierry Deutsch – CEA, France
  • Dr. ivan Duchemin – CEA, France
  • Dr. Ivan Kondov – Karlsruhe Institute of Technology, Steinbuch Centre for Computing, Karlsruhe, Germany
  • Dr. Jussi Tella – CSC-IT , Software and data solutions, Espoo, Finland
  • Postdoc Jeroen Van der Holst – University of Mons, Chemistry of Novel materials, Mons, Belgium
  • Prof. Wolfgang Wenzel – Karlsruhe Institute of Technology, Institute of Nanotechnology, Karlsruhe, Germany

Organic light-emitting diodes (OLEDs) are a relatively new technology, which provide a low-cost and potentially high-efficiency alternative to present inorganic lighting and display applications. A typical OLED consists of one layer or a stack of multiple layers of organic semiconducting material sandwiched between two electrodes. Due to the energetic disorder present in organic media, charge carriers are transported by means of hopping between neighbouring molecules and/or segments of polymers. The energetic levels of these hopping sites are disordered and often assumed to be distributed according to a Gaussian density of states. An OLED works as follows. Electrons and holes are injected in the organic material. They are transported under the influence of an applied bias voltage and their mutual Coulombic interactions either to the opposite collecting electrodes or to each other. Once an electron and a hole meet each other, they recombine to form a bound electron-hole pair (exciton) which can decay radiatively under the emission of a photon. Despite the growing commercial success of OLEDs, knowledge of many aspects of their functioning is still fragmental. To increase our theoretical understanding of the transport properties, we integrate several different simulation methods covering one or more length- and/or time-scales. This multi-scale modelling scheme consists of the following steps: 1. By means of coarse-grained Monte Carlo simulations and employing well established atomic force field parameterizations, the morphology of polymers and organic molecules can be simulated. 2. These morphologies are used in density functional theory simulation programs to calculate charge transfer integrals between sites and the energetic landscape 3. The predicted morphologies and the corresponding charge transfer integrals and energetic landscapes are used in a kinetic Monte Carlo program to simulate typical relevant transport properties like the three-dimensional charge carrier current density and recombination statistics. The described simulation scheme will be used in this project for a multi-scale modelling study of organic devices with tris(8-hydroxyquinolinato)aluminium (Alq3) as an active component. Special focus will be devoted to the influence on charge carrier mobilities of energetic and positional disorder as well as charge density. The work will be extended to model prototypical host:guest systems, where the host is a wide gap molecular semiconductor, such as 4,4′-bis(N-carbazolyl)-1,1′-biphenyl (CBP), N,N′-dicarbazolyl-3,5-benzene (mCP), or Alq3, and where the guest is the well-known triplet emitter fac-tris(2-phenylpyridine) iridium [Ir(ppy)3]. The project requires considerable and varied computational resources, in terms of cpu times, disk space and dynamic memory allocation, which fully justifies application to PRACE.

Project Title: Ab initio thermoelectrical properties of advanced nanomaterials

Project Leader: Prof. Matthieu Verstraete, University of Liege, Physics, Liege, Belgium

Resource Awarded: 3 920 000 core hours on EPCC – HeCToR XE8 and SURFSARA – Huygens P7


  • Prof. Philippe Ghosez – University of Liege, Physics, Liege, Belgium
  • Prof. Jean-Yves Raty – University of Liege, Physics, Liege, Belgium

The identification of alternative and renewable sources of energy is one of the most important challenges modern society faces, and has become more urgent and intense in the past few years. One of the most promising technologies is that of thermoelectric (TE) devices, which allow one to transform heat into electrical energy (an alternative energy source) or vice-versa (for refrigeration or heating). For the moment TE applications are however restricted to niche products in domains such as aerospace, the military, or automobile industries. Several basic technological problems still need to be solved before thermoelectrics become a competitive energy source. In particular, the efficiency of thermoelectric materials themselves will have to be roughly doubled before cost-effective, large-scale applications can be envisaged. New perspectives on thermoelectrics have been opened recently by their structuring on the nanoscale. This has allowed experimentalists to obtain impressive efficiencies in thin film samples in the lab, but optimizing these materials and transferring these new ideas to a nanostructured bulk material suitable for industrial environments and mass production remains a challenge.

Since the beginning of the current revolution in thermoelectrics, theoreticians (in particular GD Mahan and MS Dresselhaus) have played a central role, proposing new material and device paradigms, and explaining how the intrinsic limits of bulk thermoelectric materials can be overcome or bypassed. All these theoretical works have however been performed at a semi-empirical level and typically include severe approximations such as a constant relaxation time for electrons and for phonons. No group to date has yet concentrated all of the tools necessary to achieve predictive characterization of thermoelectrical materials entirely from first-principles calculations.

This application unites the competencies of the three groups performing theoretical materials science in the Physics Department of the University of Liège, in order to achieve first-principles predictions of the thermoelectrical properties of materials. The groups of Ph. Ghosez, M. Verstraete and J.-Y. Raty have complementary expertises, both in software development and in materials modelling, for semiconductor, metallic, intermetallic, and nanostructured systems; they have particular competencies in the domains of the ab initio prediction of vibrational properties, as well as electron-phonon coupling, both of which are central to thermoelectric properties. On the one hand, a fundamental methodology has been developed for the first principles prediction of thermoelectric qualities of materials within density functional theory (led by M. Verstraete). On the other hand, the methodology will be applied to two important classes of materials (functional oxides, led by Ph. Ghosez ; and intermetallics, led by J.-Y. Raty) which appear as the most promising candidates to address respectively high (600-700K) and room temperature (300-400K) thermoelectric applications.

Project Title: Precise quantum chemical calculations for methylidynium (CH+ )

Project Leader: Prof. Jacek Komasa, Adam Mickiewicz University, Faculty of Chemistry, Poznan, Poland

Resource Awarded: 3 565 044 core hours on EPCC – HeCToR XE7


  • Piotr Kopta – PSNC, Applications Department, Poznan, Poland

Advances in quantum theory and high precision computational methods lead to more and more precise predictions of properties for light atoms and molecules. A fundamental goal of such methods is to find precise solutions of the Schrödinger equation. The most precise methods employed to solve the Schrödinger equation are based on the variational principle and take explicitly into account the electron correlations in the construction of a trial wave function. The method of explicitly correlated Gaussian (ECG) functions is one of the best suited to this purpose. It has been widely used for many light atoms and molecules and has proven its ability to supply the benchmark quality results. The goal of this project is a generation of well optimized ECG wave functions for a six-electron molecule (methylidynium, CH+) at the equilibrium distance. This cation is one of the first species discovered in the interstellar space and being essential in formation of carbon hydrates. Our project aims at providing reliable data for solving questions pertaining to astrophysics and for understanding the role of CH in the interstellar chemistry. For that purpose, we plan to employ the ECG method and the massive computations to a large scale deterministic optimization of the energy and wave function according to a well-defined strategy.

Project Title: Titanium Dioxide Interfaces: Engineering Photocatalytic Activity in Mixed TiO2

Project Leader: Dr. Michael Nolan, Tyndall National Institute, Cork, Ireland

Resource Awarded: 2 746 483 core hours on UHEM – UYBHM

The TiO2-Interfaces project will undertake first principles density functional theory (DFT) simulations of technologically important interfaces between the two dominant polymorphs of titanium dioxide (TiO2), namely anatase and rutile. This is a key materials system in photocatalysis, which has the potential to be an extremely important technology for energy generation and depollution, using only sunlight and cheap, widely available materials in the process. Photocatalytically active TiO2 is a mixture of these polymorphs and understanding how the formation of the interface and associated properties determines the photocatalytic activity is necessary, which can now be performed reliably with DFT simulations. The simulations will be of models of the rutile-anatase interface and will allow us to obtain a detailed understanding of how interface properties determine the photocatalytic activity. In addition, we will examine how the presence of defects and doping can be used to modulate the light absorption properties and charge transport. Comparison will be made with the individual polymorphs to investigate how the interface affects the energetics of defect formation and doping. The results of the project will be important for future experimental work using mixed TiO2 as a photocatalytically active material.

Plasma & Particle Physics Projects

Project Title: Global electromagnetic gyrokinetic simulation in 3D equilibria

Project Leader: Dr. Ralf Kleiber, Max Planck Institute of Plasma Physics(IPP), Greifswald, Germany

Resource Awarded: 3 754 968 core hours on FZJ – JuRoPA


  • Dr. E. Sánchez Gonzáles – CIEMAT para Fusión, Spain

It is necessary to accompany stellarator experiments, as e.g. Wendelstein 7-X, by simulations of plasma microinstabilities and related turbulence. Especially important for the comparison of experiment and theory are full torus simulations for stellarator configurations since they need no approximations regarding the geometry. Gyrokinetics as a first principle based theory is well suited to describe the relevant physics. An established and flexible method for solving the gyrokinetic system of equations is the simulation via the particle-in-cell (PIC) Monte-Carlo method. For this purpose the EUTERPE code has been developed which solves the gyrokinetic equation globally in arbitrary stellarator geometry including kinetic electrons, and electromagnetic perturbations. The full kinetic treatment of the electrons allows the investigation of trapped electron effects (e.g. TEM) and the inclusion of electromagnetic effects establishes the connection to magnetohydrodynamics (MHD). By using a third species the destabilisation of MHD modes (e.g. TAE, HAE) can also be studied. These developments will make EUTERPE the first code worldwide that is able to simulate global gyrokinetic electromagnetic instabilities in three dimensions.

Project Title: Precision calculations with realistic Wave Functions of NUCle

Project Leader: Claudio Ciofi degli Atti, University of Perugia & INFN, Italy

Resource Awarded: 1 000 000 core hours on CSCS – Rosa and RZG – Genius


  • Massimiliano Alvioli – University of Perugia & INFN, Italy
  • Dr Chiara Benedetta Mezzetti – University of Perugia & INFN, Italy

Atomic nuclei, hadrons and various high density systems, like e.g. neutron stars and QCD matter at extreme conditions, are systems governed by the strong interaction. Hadronic physics is therefore that field of research which ranges from the study of the structure of nucleons and nuclei to the structure of compact high density systems in the cosmos, with resulting relevant implications of astrophysical and cosmological natures. The nature of QCD, which governs the interactions between quark and gluons confined in hadrons, is such that there is a residual interaction between nucleons, known as strong interaction. Atomic nuclei are bound by such an interaction. The quantitative investigation of description of nuclear structure at nucleon level is based upon complicated, phenomenological Nucleon-Nucleon (NN) potentials, which give rise to wave functions characterized by a peculiar short-range structure described by the so-called Short-Range Correlations (SRC) among clusters of nucleons. At short distances, the strong interaction becomes very repulsive and appreciably state-dependent, and the motion of two nucleons strongly deviates from an independent motion, and, as consequence, high momentum components are generated in the nuclear wave function. In the past the observation of SRC has been impossible due to limited capabilities of experimental techniques, and their very existence was questionable. During the last few years, modern technology allowed to overcome technical difficulties and dedicated experiments were successfully performed which clearly showed the existence and relevance of SRCs. The local density of an NN correlated pair may be up to ~7 times the average usual nuclear density, and thus comparable with densities in the core of a neutron star. The study of SRCs is therefore not only interesting per se, but also for the study of other phenomena such as formation of neutron stars, in-medium modification of nucleon properties, and the transition from the nucleonic to the underlying QCD degrees of freedom. Achievement of these goals is a difficult task since it involves the solution of the many-body problem governed by the complicated NN interaction. However, obtaining nuclear wave functions represents a fundamental goal of nowadays hadronic physics, since only by their use reliable description of the basic aspects of hadronic physics, ranging from the structure and interactions of nucleons, nuclei and compact high-density objects can be achieved. The aim of the project is the description of various phenomena involving nucleons and nuclei by using realistic wave functions of few- and many-body systems.