Pokročilé metody molekulových simulací a jejich aplikace na biologicky relevantní systémy

• Název práce v češtině: Pokročilé metody molekulových simulací a jejich aplikace na biologicky relevantní systémy
• Název v anglickém jazyce: Advanced Molecular Simulation Methods and Their Use for Biologically Relevant Systems
• Akademický rok vypsání: 2024/2025
• Typ práce: disertační práce
• Ústav: Katedra chemické fyziky a optiky (32-KCHFO)
• Vedoucí / školitel: prof. Mgr. Pavel Jungwirth, CSc., DSc.

Zásady pro vypracování

Efficient yet accurate theoretical description of molecular systems, including those of biological relevance, is
a subject of an ongoing intensive research, paving the path towards the next generation of simulation techniques.
Due to the technological limitations, today’s methods are trapped from both ends of the spectrum
– ab initio approaches suffer from poor computational efficiency, while classical approximations are limited by
their accuracy and detail of the insights provided.
In the recent years, the concept of electronic continuum correction (ECC) was proposed [1] as a promising
possibility to enhance the well-established methods of classical, force field molecular dynamics (FFMD) via
physically sound [2] charge scaling of atomic species. ECC introduces electronic polarization in a mean-field
way through scaling charges by the reciprocal of the square root of the high-frequency dielectric constant
of the solvent environment. This allows for effective charge shielding which is otherwise missing in the standard
FFMD framework. Such correction could serve as a bridge between the inherently macroscopic and microscopic
features of large-scale molecular simulations.
On the other rapidly growing front, machine learning (ML) models, e.g. neural networks [3, 4] or Bayesian
models [5], have recently been introduced to ab initio methods. Such protocols have the potential to enhance
computational efficiency of ab initio approaches while retaining as much accuracy and quantum insight into
the simulated systems as practically possible. Machine learning thus opens the doors for deep theoretical insight
into more complex molecular systems. This can subsequently be utilized for efficient development and validation
of models for large-scale, potentially organic systems.
The goal of the present thesis is to push the boundaries of molecular simulations from the bottom up,
i.e. improving ab initio approaches through machine learning protocols and similar techniques. These advanced
molecular simulation methods will be utilized towards an in-depth investigation of the nature of biologically
relevant systems harboring charged moieties, including reactive systems associated with phenomena we observe
in nature and experiments. The scope of our application aim lies within the very building blocks of life, mainly
amino acids, nucleobases, peptide residues and their relation to the surrounding solvent containing ionic species.
This will among other provide crucial guidance and validation to the quest of correct development of the ECC
framework, which enhances classical simulations of even larger, more complex molecular systems.

Seznam odborné literatury

[1] B. J. Kirby and P. Junwirth. Charge Scaling Manifesto: A Way of Reconciling the Inherently Macroscopic
and Microscopic Natures of Molecular Simulations. J. Phys. Chem. Lett. 2019 10(23), 7531-7536.
[2] V. Kostal, P. Jungwirth, and Hector Martinez-Seara. Nonaqueous Ion Pairing Exemplifies the Case for
Including Electronic Polarization in Molecular Dynamics Simulations. J. Phys. Chem. Lett. 2023 14(39),
8691-8696.
[3] C. Schran, F. L. Thiemann, P. Rowe, E. A. M¨uller, O. Marsalek, and A. Michaelides. Machine Learning
Potentials for Complex Aqueous Systems Made Simple. Proc. Natl. Acad. Sci. U.S.A. 2021 118(38),
e2110077118.
[4] K. Brezina, H. Beck, O. Marsalek. Reducing the Cost of Neural Network Potential Generation for Reactive
Molecular Systems. J. Chem. Theory Comput. 2023 19(19), 6589-6604.
[5] R. V. Krems. Bayesian Machine Learning for Quantum Molecular Dynamics. Phys. Chem. Chem. Phys.
2019 21, 13392-13410.