|
|
Quantitative structure-activity relationship and machine learning
Nierostek, Jakub ; Uhlík, Filip (advisor) ; Svozil, Daniel (referee)
Quantitative structure-activity relationship (QSAR) computational methods allow us to examine the relationship between the chemical structure of molecules and their chemical or biological properties. For QSAR calculations, widely used machine learning methods, such as deep learning models, can be used. In this work, we construct a pipeline for training QSAR machine-learning models that can predict molecular toxicity. Furthermore, we investigate the effect of molecular representation on model performance. Both our deep learning mod- els and traditional machine learning models are employed on Tox21 and Ames Mutagenicity datasets. Their performance is evaluated against recently published models for toxicity prediction using the AUC-ROC metric and, regarding certain toxicity targets, shows improvement over these models. Keywords: QSAR, machine learning, deep learning, molecular descriptors 1
|
|
|
Endohedral fullerenes: From exotic chemical bonding to molecular electronics
Jaroš, Adam ; Straka, Michal (advisor) ; Uhlík, Filip (referee) ; Patzschke, Michael (referee)
This work focuses on the theoretical study of the physical and chemical properties of endohedral fullerenes, which can be used either to study exotic chemical bond- ing or as components in molecular electronics. Part of this work is focused on the study of interactions of simple molecules enclosed in a fullerene cage, such as H2O, HF, NH3, or their dimers, and the observation of the so-called charge-shift bonding. We have also investigated the bonding interactions between actinide atoms in fullerenes and described general trends associated with this bonding. In the course of studying actinide endohedral systems, we have proposed and cali- brated a methodological approach to investigate the actinide-actinide bonding, applicable in future research. The acquired knowledge about the interactions in endohedral fullerenes al- lowed us to design molecular components usable in so-called non-von Neumann architectures. First, we published a proof-of-concept study on diatomic MX molecules (M = hydrogen or metal, X = halogen or chalcogen) enclosed in a C70 cage. The molecule enclosed inside the cage can be rotated by an external electric field; thus the whole system acts as a switch. We found that the conductivity of the system depends on the rotational state of the molecule enclosed within the system, and the...
|
|
|
Computer Simulation Methods of Soft Matter
Kazakov, Aleksandr ; Uhlík, Filip (advisor) ; Bačová, Petra (referee) ; Jirsák, Jan (referee)
iii CHARLES UNIVERSITY Faculty of Science Department of Physical and Macromolecular Chemistry Doctoral Thesis Computer Simulation Methods of Soft Matter by MS. Aleksandr D. KAZAKOV Abstract This thesis is dedicated to the theoretical study of hydrogels under different con- ditions, and implementing new simulation approaches: for polymer- and carbon- based systems. The hydrogel was modeled with presence of multivalent ions at good solvent conditions and with presence of monovalent ions at different salt concentrations and different pH pK values at poor solvent conditions. The competition between elec- trostatic repulsion with short-range hydrophobic interactions in case of pH-respon- sive polyelectrolyte lead to first-order transition in the system. The study has shown how to utilize and to tune the transition with respect to parameters. A novel hybrid MC-SCF method was implemented, which combines a mean- field representation by the machinery of Scheutjens - Fleer self-consistent field and a coarse-grained model driven by the Monte Carlo method. The study provides a comparison of the novel method with pure coarse-grained and mean-field methods. The family of fullerenes with an even number of carbon atoms n = 20-80 was calculated using the DFT method and subsequently trained using a convolutional neural network...
|
|
|
Theoretical study of branched polyelectrolytes
Prokacheva, Varvara ; Uhlík, Filip (advisor) ; Kramarenko, Elena (referee) ; Tuinier, Remco (referee)
ii CHARLES UNIVERSITY Faculty of Science Department of Physical and Macromolecular Chemistry Doctoral Thesis Theoretical Study of Branched Polyelectrolytes by Varvara PROKACHEVA Abstract This thesis is dedicated to the theoretical study of branched pH-responsive polyelec- trolytes including stars, molecular brushes as well as hydrogels in dilute aqueous salt solutions. It was investigated how external parameters, such as ionic strength, pH, and solvent quality affect the properties and the internal structure of those poly- electrolytes. The study was carried out by the development of an analytical theory which results were validated by the numerical Scheutjens-Fleer self-consistent field method as well as Monte Carlo and molecular dynamics simulations. A competition between electrostatic repulsions with short-range solvophobic in- teractions leads to complex patterns within intramolecular self-organization of bran- ched polyelectrolytes. That results in the formation of domain structures, in partic- ular, core-shell and pearl necklace structures in stars and molecular brushes, respec- tively. These domains are formed by almost neutral collapsed chains and stabilized by stretched and ionized chains. This study is relevant both in terms of creating new functional polymer materials for a wide range of...
|
|
| |
|
|
Theoretical Investigation of ethanol dehydration catalyzed by acid sites in zeolites
Vacek, Jaroslav ; Nachtigall, Petr (advisor) ; Uhlík, Filip (referee)
Zeolites are a group of aluminosilicate minerals with catalytic properties. They may be used for many industrial applications such as catalytic cracking of oil. Zeolites are also capable of converting ethanol to diethylether and ethylen. This reaction is known as dehydration of ethanol. The reaction is potentially interesting as a way of converting ethanol to more valuable molecules. An experimental study (Shashikant A. Kadam, Mariya V. Shamzhy, 2018) has proven that diethylether is the preferred product when the temperatures are low and the partial pressure of ethanol is high. Ethylen is more significant product with higher temperature and lower partial pressure of ethanol. Aim of this thesis is to determine the mechanism of dehydration of ethanol. Furthermore it was attempted to explain the behavior of the reaction under different circumstances. The research was done in silico using the methods of computational chemistry. Such methods give information on the geometry and the energy of systems of molecules. Thus computational chemistry can be used to investigate the relational path and activation energy of the studied reaction. This thesis is a theoretical study of dehydration of ethanol catalysed by a zeolite.
|
|
|
Počítačové modelování větvených polymerů
Preisler, Zdeněk ; Uhlík, Filip (advisor) ; Netopilík, Miloš (referee)
In this work we study properties of branched polymers in a good solvent. We focus on problematic related to the size exclusion chromatography and predicting elution behavior of randomly branched polymers. We developed a software for generating self-avoiding walks (SAW) of any given non-looping architecture on a cubic lattice using Monte Carlo (MC) simulation and vali- date its reliability by presenting the scaling of different architectures: linear, 3-arm star and 6-arm star and asymmetric star. We calculate distribu- tion coefficients and calibration curves for size exclusion chromatography for various architectures to validate that the hydrodynamic radius is more suitable for predicting elution volume than the radius of gyration. Then we propose a new method for, although approximate, a very fast estimation of radius of gyration and hydrodynamic radius for different architecture using a graph method. It is done by comparing MC results with results obtained from graph theory. Then we introduce a correction to graph-theory results to fit the MC. At the end we present depletion layer calculation from MC and self-consistent field (SCF) method for polymers and their comparison. We show how calculation of depletion layer using SCF can be improved to get significantly better agreement with MC results. v
|
|
|
Computer Modelling of Macromolecules
Khirnova, Oleksandra ; Uhlík, Filip (advisor) ; Jirsák, Jan (referee)
In the present work we study diffusion of linear and cyclic polymers. We run simulations using Brownian dynamics in θ-solvent and good solvent with several hydrodynamic models. Obtained results were compared with existing theories. In the present work we study diffusion of linear and cyclic polymers. We run simulations using Brownian dynamics in θ-solvent and good solvent with several hydrodynamic models. Obtained results were compared with existing theories. In the present work we study diffusion of linear and cyclic polymers. We run simulations using Brownian dynamics in θ-solvent and good solvent with several hydrodynamic models. Obtained results were compared with existing theories. In the present work we study diffusion of linear and cyclic polymers. We run simulations using Brownian dynamics in θ-solvent and good solvent with several hydrodynamic models. Obtained results were compared with existing theories. In the present work we study diffusion of linear and cyclic polymers. We run simulations using Brownian dynamics in θ-solvent and good solvent with several hydrodynamic models. Obtained results were compared with existing theories.
|
|
|
Computer study of protein folding using simplified models
Nierostek, Jakub ; Uhlík, Filip (advisor) ; Jirsák, Jan (referee)
The aim of this work is to design and describe a suitable coarse-grained protein model, on the basis of which protein-folding will be studied. The model will be implemented as a computer program, development of the model in time will be simulated by Hamiltonian Monte Carlo. Using computer simulations, not only the protein-folding itself will be investigated, but also the quantities that characterize the process and the similarity of the real and simulated protein's native conformation. Keywords: protein folding, computer simulation, Hamiltonian Monte Carlo, coarse-grained model 1
|
|
| |
guest ::
