ZOZNAM ÚSPEŠNÝCH PROJEKTOV VO VÝZVE 2/24:
Molecular dynamics computer simulations as a supporting method for AFM imaging
AUTOR: Dušan Račko
The project is aimed on combining coarse grained molecular dynamic simulations of catenated DNA molecules with the experimental AFM imaging. The simulations will not only help understanding interactions of DNA with mica surface, but after the model is developed the simulations will provide a straightforward method to generate huge amounts of AFM images. These images in turn could be used to train algorithm in the AFM imaging pipeline that are used for automated detection of DNA topology. Currently, experimental structures in a range up to 100 molecules were prepared. The work is done in collaboration with Pyne’s lab and the University of Sheffield.
Transcriptomics analysis of tumor breast cancer tissue using next-generation sequencing and clinical data
AUTOR: Michal Gala
Breast cancer (BC) is heterogeneous and complex disease that needs accurate diagnostics ensure appropriate treatment. To achieve this goal, in recent years, whole transcriptome sequencing groves in popularity. Here, we plan to use Devana computational resources to process raw RNA sequencing reads to obtain whole transcriptomics profiles of patients with breast cancer. The processing pipeline of next-generation sequencing (NGS) RNA data involves several steps that depend on the specific methods of sequencing the RNA sample and on the specific computational tools used for processing raw reads. The process itself can be divided to several steps including quality check of reads, adapters cutting, deduplication, aligning reads to reference genome and the final step, counting number of reads for each transcript. All these steps are essential parts of RNA sequencing data processing. Due to natural size of human genome, which is about 3.4 gigabases and the large amounts of data produced in NGS, counted in tens of millions of reads, these tasks cannot be processed on personal computer and required high performing computing resources. The final output – raw counts gene matrix represent essential input for various task such as BC subtype predicting, predicting of risk of recurrence or identification of prognostic and predictive biomarker signature to optimize treatment for early breast cancer patients.
Molecular magnets for quantum technologies
AUTOR: Erik Čižmár
The project’s primary objective is to understand the influence of molecular design on magnetic anisotropy and exchange interactions, which affect quantum coherence and quantum entanglement in molecular magnets. The knowledge of the mechanisms of quantum coherence and quantum entanglement in molecular magnets is crucial for the promising application of molecular magnets in quantum information processing and quantum sensing. Nowadays, quantum chemical approaches, such as multireference ab initio methods as implemented in ORCA or MOLCAS packages, allow us to predict and study the properties of molecular magnets. Unfortunately, for large molecules with more than 120 atoms, the computational costs exceed the possibilities of small computational workstations. Within this project, we propose to study the influence of molecular structure on the magnetic anisotropy of a series of tetranuclear molecular spin clusters consisting of transition metal ions and/or lanthanide ions, which are suitable for further experimental and theoretical study of quantum coherence and quantum entanglement, using a multireference ab initio approach (SA-CASSCF/NEVPT2 or SA-CASSCF/DLPNO-NEVPT2) as implemented in the quantum chemical package ORCA.
Theoretical predictions and synthesis of high-entropy diboride systems with different molar ratios of transition metals
AUTOR: Inga Zhukova
High-entropy materials have attracted considerable interest due to their unique, improved properties and large configurational entropy. Out of these, high-entropy ceramics (HECs) are of particular interest since the independent solubility of cations and anions results in increased configurational entropy. Forming a system of five transition metals (Hf, Nb, Ta, Ti, Zr) and boron atoms (B2) represents a promising material, which shows crystalline phase stability, high strength, and thermal oxidation resistance under extreme conditions. The present work is dedicated to the theoretical aspects, and the predictions of diboride structures with different molar ratios of transition metals. The paper highlights the special quasirandom method (SQS), calculations of lattice parameters, energetic properties and different molar ratios on atoms. Three most promising selected compositions were successfully synthesized via boro/carbothermal reduction synthesis, and the experimental results were compared to the theoretical data.
Rare-Earth Oxides on W-Al2O3 Electrode Analysis
AUTOR: Eva Pospisilova
There are two main challenges which I would like to address within my one-year post-doc at the Institute of Materials and Machine Mechanics, SAS, Bratislava.
The first task concerns identification of La oxides emerging on tungsten electrode alloyed with 2 wt. % La2O3 and sealed in Cu matrix during welding in ambient air, determination of its properties such as the most stable structure both at room temperature and under working conditions (> 1400C), magnetic state, work function and heat conductivity. Various oxides, e.g. La2CuO4, La6W2O15, La2WO6, LaCuO3, LaCuO2, CuWO4, Cu2WO4, Cu3WO6, prevent degradation of W electrode in air by forming a heap atop melted 2mm-thick layer of Cu electrode matrix, thus hindering further evaporation of Cu and concomitant exposure of W electrode to air. Because the main role of Cu is to conduct heat away, we are naturally interested in the heat conductivity of the covering oxides and in their work function as well.
This query would be tackled via standard DFT setup using VASP, enabling to calculate both phase diagrams and work function. As for magnetism and strong interelectron interaction posing well known problems for cuprates and tungstates, Hubbard U-correction to DFT PAW GGA/PBE or SCAN functional would be attempted. Heat conductivity shall be handled with Alamode software.
The second task deals with replacement of rare and/or expensive elements that are nowadays imported prevalently from Russia and China, such as scandium (Sc), in Al-based alloys used for Additive Manufacturing. The role of Sc in alloys for 3D printing is to strenghten the material while rendering it ductile and crack resistant. Including other rare and expensive elements of the same or neighbouring groups such as Zr, Y, Yb, there is not a single element suitable for this purpose, since Sc has the unique property to substitute Al in its face-centered cubic lattice for coherent Al3Sc.cP4 precipitate phase almost without any mismatch. Hence, we are to replace Sc with some combination of atoms with similar diameters slightly above and below Sc, e.g. Ti and Zr. Mechanical properties (tear-resistance tests) and temperature stability of such surrogate precipitate phases will be immediately sought after.
As concerns zero and finite-temperature stability, again, standard DFT setup will be employed. As for simulations of scale precipitate models embedded in Al matrix, an application of Machine Learning Force Fields (MLFF) as already implemented in VASP code is intended. MLFF would be trained on small (toy) models of precipitates and subsequently introduced to much larger samples. In addition, MLFF facilitates meso-scale or multi-scale modelling of complex interactions of precipitates with dislocations, which are essential to enhancement of material strength.
Development of numerical modelling of seismic motion and seismic ambient noise
AUTOR: Jozef Kristek
One of the important factors contributing to the minimization of earthquake damage is a sufficiently accurate estimation of the seismic motion during future earthquakes. Seismic motion can be anomalously amplified or prolonged by surface sedimentary structures. In the absence of seismic observations, numerical modeling is an irreplaceable tool to better understand the conditions determining seismic motion and to improve the estimation of seismic motion. Given the evolution of computational resources, the increasing demands for model accuracy and frequency range, it is essential to adequately develop sufficiently accurate and efficient numerical methods. Therefore, the project is primarily focused on development of numerical methods for modelling of seismic motion. The project will also address fault rupture propagation and seismic noise. The modelling of these processes is computationally very demanding and therefore requires a specific approach.
Circular chromatic index of some regular graphs classes
AUTOR: Dušan Bernát
There are several graph theory conjectures concerning circular chromatic index of snarks or other regular graphs. To prove or disprove such a claim usually all graphs of given class (e.g. with given number of vertices, or given girth, etc.) must be examined. We already have computed circular chromatic index for all snarks of order 30 with girth at least 5. The aim of current work is to extend the results to snarks having girth at least 4, and to snarks of higher order, particularly 32, 34, 36. This would lead to establishing new bounds of chromatic index or validation of some theorems also for these graphs.
Using deep learning approach for automated segmentation of knee articular cartilages in 3D MRI data of human patients
AUTOR: Andrej Krafčík
Introduction: To morphologically describe, segment and quantify structures in volumetric medical data. The size of data in current 3D MRI is extremely large, what results in enormous time and effort needed by certified radiologist to make a diagnostics. To speed up, automate and objectify of this process, is therefore actually very desired. Strong tool for fulfillment of these requirements is application of artificial intelligence approaches, in particularly deep learning methods as neural networks, while high-quality manually segmented data, which enters into neural network training, validating and testing process, will be also used as a reference.
Objective: Therefore in our standard project, we would like to focus on training of our convolutional neural network (CNN) with 3D U-Net architecture, as the method of artificial intelligence, onto manually segmented cartilages of human knee-joint MRI big data with high resolution obtained by 3-dimensional double-echo steady state (3D-DESS) MRI sequence, via using hardware resources of high performance computer (HPC) Devana.
Aims: The first aim is to estimate number of required filters in each resolution stage of analysis and synthesis paths of our 3D U-Net CNN architecture (which defines number of trainable parameters) for sufficient CNN model description necessary to correctly segment our 3D-DESS MRI dataset of human knee-joint patients with high resolution. Next step and the main aim is to train such CNN model for automated cartilage segmentation with further incorporation of such model into software-tool as the part of our solved project APVV-21-0299 (Automatic data evaluation tool from the longitudinal quantitative MRI studies of articular cartilage) focused on automated segmentation of knee articular cartilages, which would be used in long-term study of applied therapy success monitoring in human patients pathology of knee joint cartilages.
Methods: To complete both aims, the CNN with 3D U-Net architecture as deep learning artificial intelligence method in Python environment by using available module TensorFlow 2.12.0 with GPU support will be used.
Atomistic predictors of thermodynamic and transport properties at the materials interfaces
AUTOR: Andrej Antušek
Materials interfaces play a crucial role in determining the properties and performance of composite systems. These interfaces led to a wide range of applications, from advanced electronics to structural materials. At the interface, the interaction between distinct materials can lead to unique physical, chemical, and mechanical properties that are not present in the bulk phases. We will extend our previous research addressing thermodynamics and transport properties of Ag/AlN and Au/AlN interfaces. We will use density functional theory to investigate vacancy formation energies and vacancy migration barriers at the interfaces. The first principles results will be presented together with experimental results provided by EMPA Dubendorf.
Development of predictive models for NMR shielding of ions in liquids. Part II
AUTOR: Andrej Hurajt
Accurate ab initio NMR shielding constants are vital for establishing absolute shielding scales and refining nuclear magnetic dipole moments. These refinements enable direct NMR shielding measurements, reducing reliance on chemical shifts and lowering experimental costs. Recent contributions to Stone’s Nuclear Data Tables reflect our progress in this area.
Collaborating with CERN ISOLDE, we achieved unprecedented accuracy in measuring nuclear magnetic dipole moments of beta-decaying nuclei using beta-NMR techniques, crucially supported by NMR shielding calculations.
We previously developed models for NMR shielding of beta-NMR probe ions in EMIM-DCA ionic liquid, identifying limitations in classical force field methods.
To overcome these limitations, we will use machine learning (ML) potentials in molecular dynamics simulations. Utilizing VASP-implemented ML potentials, we aim to refine the solvation structure modeling of Tl+ ions in water and beta-NMR probes in ionic liquids.
Simulácia defektov štruktúr v hybridných Peroskitoch pomocou DFT metód
AUTOR: Róbert Turanský
Cieľom projektu je skúmanie/simulácia vlastnosti hybridných peroskitov (PV) použitím rôznych DFT (VASP, CPMD, CP2K) metód. Hybridné perovskity (PV) ako optické absorbéry vo fotovoltaických článkoch by mohli mať teoretickú fotoelektrickú účinnosť porovnateľnú s pokročilými kremíkovými technológiami, pričom technológia prípravy takýchto solárnych článkov na báze PV by mala byť oveľa výhodnejšia. Účinnosť premeny svetla na fotoprúd však môže byť ovplyvnená výskytom rôznych defektných stavov v štruktúre PV filmu a tiež tepelnou závislosťou optickej absorpcie.
Cieľom projektu bude štúdium vplyvu základných typov defektov – vakancií na elektronickú pásovú štruktúru MAPbI3, (Br3,Cl3), dielektrické funkcie a optické absorpčné prechody pomocou DFT metód (DFT a hybridnej DFT). V ďalšej časti sa projekt zameria na opis jemnej štruktúry hladín v blízkosti vodivostného pásu a na vplyv tepelných vibrácií mriežky vyvolaných fónickými módmi na optickú absorpčnú hranu PV. Pri štúdiu dynamiky mriežky sa predpokladá simulácia pomocou DFT a elektronické hladiny ležiace v blízkosti vodivostného pásma PV pomocou post-DFT (TD-DFT, G0W0, BSE). Simulácie kľúčových elektronických a fotoelektrických parametrov poskytnú modelovú/teoretickú/ podporu experimentálnej časti, ktorá sa zameriava na pasiváciu defektov počas rastu PV vrstiev + ladenie optoelektronických vlastností.
First-principles calculations of electronic effects in 2D materials and van der Waals heterostructures relevant for future spintronics devices
AUTOR: Timon Moško
The project aims to explore the electronic structure of two-dimensional ma- terials and van der Waals heterostructures, with a focus on their relevance to spintronics applications. Our approach involves conducting large-scale first- principles calculations based on density functional theory to examine electronic and thermal transport properties, charge-to-spin interconversion efficiencies, and spin-orbit torque in relevant 2D magnetic, altermagnetic, and supercon- ducting materials, as well as their van der Waals heterostructures. We will specifically investigate newly synthesized 2D magnetic metal iodides, interca- lated transition metal monolayers, altermagnetic materials, and superconduc- tors with non-trivial pairing mechanisms. This project’s anticipated outcome is identifying promising materials and heterostructures for next-generation elec- tronic and spintronic devices.
In silico study of the decomposition pathways of pesticides in the atmosphere
AUTOR: Ivan Černušák
The objective of this project is to provide the thermo-kinetic data on the atmospheric reactivity of pesticides present in environment. The main objective is to improve the understanding of the homogeneous reactivity in gaseous phase of selected Volatile Organic Compounds (VOC) with major atmospheric photo-oxidants such as OH. The complementary approaches will be used to better address the lack of data in the field of atmospheric chemistry and to provide a set of reliable kinetic and mechanistic data in order to improve the relevance and accuracy of dispersion software or climate models.
Adapting State-of-the-Art Atmospheric Science Deep Learning Models for Slovak Weather Conditions
AUTOR: Peter Pavlík
The massive amounts of Earth observation data collected over the years make meteorology a good match for data-hungry deep learning approaches. In this interdisciplinary project, we want to focus on developing deep learning models for short-term weather forecasting (precipitation nowcasting) and energy meteorology (PV output prediction). Deep learning models provide a way to parametrize the subgrid-scale processes and learn the patterns in the data. However, a significant challenge facing these models is that we can not guarantee the generalizability of the learned biases from one region to a different one where we want to apply the model. At the same time, not all regions of the world are the same in regards to the amount of Earth observation data available, making accurate models difficult to train only with local data. The primary aim of this project is to investigate whether transfer learning can help us to adapt existing state-of-the-art precipitation nowcasting and PV output prediction models trained using datasets from different geographic regions for the Slovak weather conditions.