CSC's Scientific Customer Panel selected 22 projects that will be run as CSC's Mahti Pilot Projects during the acceptance phase starting in April 2020.

The approved projects are:

Joni Vuorio, University of Helsinki: Role of SHANK3 in Ras-driven oncogenesis

SHANK3 is a scaffold protein and actin regulator in excitatory synapses. It forms a protein-protein network critical to the maintenance and plasticity of the synapse by connecting effector proteins on the plasma membrane and cytoskeleton. A recent 3D structure of SHANK3 reveals its N-terminal SPN domain to be a Ras-association domain, through which it sequesters Ras proteins and subsequently regulates the binding of other effector proteins. The Ras proteins are well-known molecular switches, activating signaling in normal tissues as well as in cancer. This project uses molecular dynamics simulations to investigate the Ras-SHANK3 interface to uncover how the binding of Ras translates to functional changes in SHANK3 in both health and disease. This will help to understand the role of SHANK3 in, e.g., the onset of Ras-driven human carcinomas. The simulation work is bridged to experiments via collaboration with top cell biologists.

Software: GROMACS

Minna Palmroth, University of Helsinki: Carrington storm in modern infrastructure

The Carrington storm, the largest space weather event in observational history occurred in 1859. Infrastructural damage was limited to the telegraph, suffering from so-called geomagnetically induced currents (GIC). There are virtually no studies outlining the impacts of the most extreme storms on modern infrastructure. The modern computational capacity is still not large enough to model the most extreme driving in 3 dimensions (3D). We propose to reconstruct the Carrington effects in a simplified 2D setup. We run two simulations, one with moderate and one with extreme driving:

  1. By comparing, we measure the magnetospheric magnetic field variations, the source of the GIC. We determine how much the levels scale from moderate to extreme driving.
  2. Similarly, we measure the scaling of precipitating particle energy (Grandin et al. 2019).

Software: Vlasiator

Xavier Prasanna Anthony Raj, University of Helsinki: Role of seipin in lipid droplet biogenesis

Lipid droplets (LDs) are organelles that serve as lipid reservoirs for membrane synthesis and energy metabolism. Abnormality in lipid homeostasis is associated with disorders such as obesity and insulin resistance. Understanding the process of LD formation has significant implications in the treatment of these disorders. Currently, it is known that LD formation occurs at the endoplasmic reticulum (ER) and is orchestrated by several proteins, the most important one being seipin. There is evidence suggesting that the formation of LDs begins with the accumulation of neutral lipids within the ER membrane to form a lens-shaped structure that seipin transforms into an LD. This picture is tempting but the molecular-scale understanding of how the process takes place is missing. Here, we use biomolecular simulations to explore the initiation and formation of LDs. The study aims to identify the key factors that regulate LD biogenesis, whose impairment could lead to metabolic disorders.

Software: GROMACS

Juho Liekkinen, University of Helsinki: Tear fluid lipid layer under non-equilibrium spreading and squeezing

One of the most common non-fatal diseases, affecting annually one third of the world's population, is the dry eye syndrome (DES). It is especially common among computer users and the elderly, with symptoms including irritation, redness, discharge, and fatigued eyes. DES results mainly from impaired tear fluid lipid layer (TFLL), which covers the human cornea. The TFLL is essential for the basic health and optical properties of the eye as it maintains the homeostasis and water evaporation from the cornea. People suffering from DES have an altered lipid composition in the TFLL, which disrupts the TFLL function. This project uses biomolecular simulations to investigate the biophysical properties of the TFLL and the role of its molecular composition in its function. In particular, we aim to unravel how a realistic mixture of TFLL lipids is able to reduce the evaporation of water from the eye surface, and how impaired TFLL with DES patients give rise to the elevated evaporation.

Software: GROMACS

Erkka Frankberg, Tampere University: Glass plasticity at room temperature

Oxide glasses are important for applications ranging from smartphone screens to window panes and they show great promise for modern electronics, including potential uses in optoelectronics, flexible electronics, photovoltaics, single-electron transistors, and battery technologies. These glasses allow for a wide range of tailored, functional properties, from full dielectrics to tuned semiconductors coupled with visible light transparency, and good chemical and thermal stability. However, in practical terms inorganic oxide glasses are considered brittle, which has led to the current design paradigm of glass and ceramic materials. We aim to transform the current paradigm by attempting to verify that a bulk oxide glass can deform plastically at room temperature.

Software: LAMMPS

Vivek Sharma, University of Helsinki: Role of protein, lipid and water dynamics in the proton pumping mechanism of respiratory complex I

The energy currency of the cell, ATP, is synthesized in the mitochondria by a chain of enzymes called respiratory complexes I-V. The electron transfer reactions by complexes I-IV generate a proton-motive force (PMF) that is coupled to the synthesis of ATP by complex V. The respiratory complex I, ca. 1 MDa in size, contributes to about 40 % of the total PMF. A high-resolution (3.2 Å) structure of complex I from yeast Yarrowia lipolytica was recently solved in our combined structural-computational work. The data revealed a novel type of lipid-protein architecture in complex I, the physiological function of which is not known. In this project, we will perform large scale atomistic molecular dynamics simulations to understand the role of lipid-protein interface in the mechanism and regulation of complex I. The results from our project will provide atomistic insight into cell respiration, and will have a far-reaching impact on health and energy sectors.

Software: GROMACS

Hannu Häkkinen, University of Jyväskylä: Photoabsorption and photoelectron effect in dynamical nanoparticles

This project will demonstrate a successful implementation of a computational scheme to simulate interactions between light and nanoscale matter, taking into account the thermal dynamics of atoms. Photoelectron and photoabsorption spectra of structurally known small silver-based nanoparticles, protected by surface ligands, will be studied by using the density functional theory and molecular dynamics methods and compared to high-quality experimental data on the particles. The results will form the basis to understand dynamical light-matter interactions in nanostructures whose electron distributions are highly heterogeneous both in real space and energy space.

Software: GPAW

Maarit Käpylä, Aalto University: How does small-scale dynamo action occur in the Sun?

We model the generation of small-scale magnetic fields by means of direct simulations of the equations of isothermal magnetohydrodynamics. We particularly address the question of small-scale field generation in systems where the smallest scales of velocity and magnetic fields are widely separated. This is the case, for example, in the Sun where the plasma velocity has much smaller structures than the magnetic field, and whose magnetic field generation mechanisms are poorly understood. This hinders the full understanding of its magnetic activity manifestations and hence the possible mitigation of their impact onto civilization. The envisaged solar-like regime is very challenging to study numerically because huge spatial resolutions are required. We attack this problem with the largest simulations of their kind in the world running with 65,536 cores and using up to 36 million CPU hours.

Software: Pencil Code

Oliver Gould, University of Helsinki: ThermOdynaMics of Cosmological phAse Transitions (TOMCAT)

Shortly after the Hot Big Bang, the universe was extremely hot and dense. As it cooled, the whole universe may have gone through a phase transition, analogous to the transition from gas to liquid. Today, the remnants of such a dramatic event may be visible as gravitational waves.

Software: susy, su2adj

Tomasz Rog, University of Helsinki: Membrane proteins in nonnative environment

Membrane proteins, including receptors, enzymes, channels, and transporters, are key players in almost all cellular processes. These include signaling, cell adhesion, energy conversion, and metabolism. They are coded by 20-30% of genes in the majority of known genomes. Membrane proteins also constitute two-thirds of proteins targeted by drugs, and they bind approximately 50% of small molecular therapeutic substances. Nonetheless, since membrane proteins are insoluble in polar solvents, they are more challenging to work with experimentally. To perform any basics studies of membrane protein, they have to be isolated from cells and then stabilized in an artificial environment. In this project, we are going to investigate the effect of non-natural environment on membrane proteins; specifically, g-protein-coupled receptors, behavior. We will investigate two frequently use assemblies stabilizing membrane proteins: detergent micelles and lipid nanodiscs.

Software: GROMACS

Pavel Buslaev & Gerrit Groenhof,University of Jyväskylä: Rotary motors: estimating ratcheting barriers in ATP synthase with atomistic moleculardynamics simulations

Molecular machines help cells to convert energy and matter to stay alive. These machines are perfect templates for bio-inspired nanotechnology, but understanding the basic mechanisms of these machines is challenging and very difficult to achieve by experiment alone. We propose to use computer simulations to investigate the Fo-F1 ATP synthase complex, for which high-resolution structures have recently become available. This membrane-bound motor converts proton (pH) gradient over the membrane into chemical energy stored in ATP molecules. The putative mechanism is that through rotation, the membrane Fo part translocates protons from the low to high pH side. Via a stalk, this rotation is coupled to conformational changes that drive the ATP synthesis in the soluble F1 region. To verify the validity of the rotation-hypothesis, we will compute the free energy barriers for the rotation of the membrane bound Fo part by means of umbrella sampling along collective angular rotation coordinates.

Software: GROMACS

Miguel Caro, Aalto University: Machine learning force field simulation of carbon nanotube growth

Carbon nanotubes have emerged as versatile materials for scientific and industrial applications, owing to their impressive strength and light-weight nature. Scientists know from experience that nanotubes can be grown by the catalytic activity of iron and other metals on carbon-based precursors.

Unfortunately, there is lack of fundamental understanding (at the atomic scale) of the growth mechanisms leading to nanotube formation and driving the observed mechanical and electronic properties. This prevents further optimization of nanotube synthesis and wider adoption of nanotube-based technologies.

In this project we will train an accurate machine-learning force field from highly accurate quantum-mechanical reference data. This new force field will allow us to model the formation of carbon nanotubes with unprecedented level of realism, a first step towards a complete understanding of carbon nanotube growth mechanisms and simulation-based optimization of growth parameters.

Software: VASP

Antti Poso, University of Eastern Finland: Hot then cold – does temperature matter for holdase activity of chaperones?

Chaperones increase the folding yields of soluble proteins by suppressing misfolding and aggregation, but how they modulate the folding of integral membrane proteins is not well understood. We will study the E. coli's chaperone SurA, responsible by the folding of β-barrel outer-membrane receptors relevant for bacterial virulence. SurA prevents proteins misfolding and agglomerating by stabilizing the nascent polypeptide. Our group has identified that SurA is heat-activated and we were able to measure the minimum temperature that acts as an activation switch. However, the conformational change induced by this process remains unelucidated. Preliminary SurA simulations in different temperatures suggest that high temperatures can lock the protein into a single conformation set, however, the inclusion of substrates and more sampling are essential to draft conclusions. Lastly, we intend to use the derived information (conformations) for drug discovery by targeting novel druggable pockets.

Software: GROMACS

Adam Foster, Aalto University: Simulated nanoendoscopy on a cell membrane

Breakthrough experimental results have applied a novel nanoendoscopy technique based on the atomic force microscope to measure living cells and their internal structures with nanometer resolution without compromising their integrity. However, a great challenge lies in the reconstruction of the real structures and dynamics of the objective from the measured AFM images. The signal is a convolution of cell dynamics, tip shape and the buffer solution, and very difficult to analyze systematically. In this project, we will combine comprehensive free energy molecular dynamics simulations to provide interpretation of the experimental results, taking advantage of the resources available to actually reach the length and timescales necessary to touch reality.

Software: LAMMPS

Kari Rummukainen, University of Helsinki: Chiral phase transition at Nf=4 QCD

Quantum chromodynamics (QCD) has a chiral phase transition at high temperatures, relevant for cosmology and probed in high-energy nucleus-nucleus collision experiments. The nature of the transition in the chiral limit, where the quarks are massless, is still under dispute. In this project we study the transition in QCD with 4 light quarks, using large scale lattice Monte Carlo simulations. Previous simulations have suffered from large lattice cutoff artefacts. Here we ameliorate them with a novel dislocation prevention algorithm, enabling us to obtain more reliable physical result than the previous simulations.

Software: lattice MC code

Rupert Gladstone, University of Lapland: COLD Antarctic ice and Southern Ocean Simulations

The Antarctic Ice Sheet is losing mass. Its margins, where ice meets ocean, are changing. Warmer waters penetrate under the floating portions of the Antarctic Ice Sheet, causing melting and thinning. Our ongoing Academy of Finland funded project "Coupled Ocean and Land ice Dynamics" (COLD) will both develop and apply existing state-of-the-art modelling tools, to investigate ice sheet – subglacial hydrology – ocean cavity interactions and patterns of change in both landward and seaward directions. This is motivated by understanding Antarctic Ice Sheet stability and its contribution to sea level. We want to utilize the pilot phase in order to produce our first simulations of the whole Antarctic ice sheet using the full-Stokes model Elmer/Ice at an unprecedented fine resolution. We will also use the ocean model ROMS to simulate the whole Southern ocean in preparation for coupled simulations.

Software: Elmer/Ice

Fredric Granberg, University of Helsinki: Quantifying irradiation Damage with quantum Accurate Machine-learning simulations

The aim of this project is to obtain atom-level representations of defects in irradiated tungsten using simulations driven by quantum-accurate machine-learning models. Tungsten is the material of choice for the wall material of fusion reactors, which will be subject to intense irradiation. Simulations of radiation damage in tungsten have been limited in accuracy by the classical interaction models. Recently, we have showed how machine learning (ML) leads to a significant boost in accuracy, making large-scale quantum-accurate simulations possible. We will use a multi-scale simulation approach to first with classical techniques produce the defect structures, then with the machine-learning model accurately relax the systems, and finally carry out RBS/C simulations to directly compare the results with experiments. This approach is unprecedented, as the ML model for tungsten has just been published, and the computational time of these simulations requires state-of-the-art computer clusters.

Software: LAMMPS

Andrea Sand, University of Helsinki: Electronic Excitations in Threshold displacement events in Tungsten

In this project, we will carry out quantum mechanical simulations of the initial stage of radiation-induced recoils near the threshold energy for defect formation in tungsten, including full atomistic dynamics and electron excitations. Such large scale computationally intensive calculations have not to date been carried out, and will provide a completely new level of insight into the quantum mechanical effects in these recoil events, which are of utmost importance to radiation damage formation in materials. The applicability and accuracy of these computational methods applied to the heavy projectiles treated in this project has been demonstrated by our recent work. With this project, we will take a first exploratory step towards building a full quantum understanding of radiation-matter interaction, with direct application relevance.

Software: Qball

Ilja Makkonen, University of Helsinki: Quantum Monte Carlo Simulation of Positrons in Defected Solids

Positron annihilation spectroscopy is used to infer atomic and electronic structure in materials. PAS measures radiation from annihilation events of electrons and the positrons in a sample.So far theoretical predictions have been provided by density functional theory, which has it's accuracy limited to the quality of the correlation functional. To overcome DFT, we developed a quantum Monte Carlo method for positron calculations to study positron annihilation in an unprecedented accuracy.

QMC method has provided very accurate descriptions of the positron-electron wavefunction. By now, we can calculate the crucial positron annihilation parameters in a sample, such as the positron lifetime and the Doppler broadening spectrum of the radiation originating from the annihilation events. The research has so far concentrated around bulk silicon, diamond, aluminium nitride and lithium, but we want to expand the studies to consider also defects in these materials.

Software: CASINO

Asier Lopez Eiguren, University of Helsinki: Gravitational Waves from Axion Strings

Dark matter constitutes 28% of the energy content of the universe, but its nature still remains unknown. Axions are predicted in a possible extension to the Standard Model of particle physics, i.e. the Peccei-Quinn (PQ) mechanism which explains why the neutron has no electric dipole moment, and are one of the most promising dark matter candidates. In the post-inflationary PQ scenario topological strings (TS) are produced and they are one of the main sources for axions. In this project we will use advanced computing techniques to study networks of axionic topological strings, which produce a gravitational wave (GW) signal. We will accurately characterise the GW power spectrum produced by such networks, which will be used to constraint the axionic dark matter scenario via current (LIGO) and future (LISA) direct GW observatories.

Software: ON-GW

Kari Laasonen, Aalto University: Hydrogen evolution on Pt

The atomic scale electrochemical reaction modelling is challenging since we need to model the catalyst surface, the water and the voltage of system. We have done several project related to Hydrogen Evolution Reaction (HER). In this project we would like to study the HER on Pt(111) water interphase at constant voltage. Pt is the best catalyst for HER and the low index Pt surfaces are the ONLY systems of which there is detailed experimental data, which make the meaningful comparisons possible. Our aim is to do best possible DFT calculations that can be compared to experiments and they will also serve a reference to more approximated methods. We need to use thermodynamical integration methods to map the reaction paths and especially the constant voltage (CV) simulations are very challenging. The CV require calculations of different size of systems.

Software: CP2K

Karoliina Honkala, University of Jyväskylä: Fluxionality of supported Pt and Rh clusters from ab initio molecular dynamics

Small supported metal clusters can have multiple low-energy structural isomers. Interconversion of isomers from one to another under the influence of high temperature and due to a changing amount of adsorbates is called fluxionality and it may influence the catalytic efficiency of a system. In the present project, we will carry out ab initio molecular dynamics (AIMD) simulations to explore the structural dynamics of zirconia-supported Pt and Rh clusters in order to understand their fluxionality. AIMD simulations will give information on cluster isomerization pathways and energetics, allow comparison of dynamic behavior of two different metals and cluster sizes, and explore the influence of adsorbed CO on structural dynamics of clusters. The obtained results will be used to derive theoretical concepts to rationalize the catalytic behavior of the studied systems.

Software: GPAW