Logo Logo
Hilfe
Kontakt
Switch language to English
Advanced techniques for the computer simulation and analysis of biomolecular systems
Advanced techniques for the computer simulation and analysis of biomolecular systems
The Helmholtz free energy is one of the central quantities of classical thermodynamics, as it governs important chemical properties such as equilibria or reaction kinetics. It is, therefore, a desirable quantity to measure, predict, and understand. Unsurprisingly, many methods exist to compute free energy differences between two states of a system. In this thesis, the density of states integration method (DSI) is developed; it detects which subsystems mainly contribute to the free energy difference. The method utilizes the velocity density of states function (VDoS) of each atom to calculate its contribution to the vibrational free energy. It is possible without any approximation to assign fractions of the vibrational free energy to meaningful subsystems, where the local free energy difference is the sum over all atoms comprising that subsystem. In this way, large local changes can be identified (free energy hot-spots), which is crucial for the understanding of free energy differences. The validity and usefulness of DSI is shown via several examples and comparison with state of the art free energy methods. In addition to the development of DSI, this thesis also focuses on free energy barriers in the context of investigating the reaction mechanism of Sirtuin~5, a lysine deacylase class~III. The relationship between the configuration of the enzyme's active site and the height of the reaction barrier is studied by computing minimal energy paths for the catalyzed reaction starting from many different (educt) configurations. Using the power of machine learning, atom-atom distances influencing the activation barrier are identified, allowing for a comprehensive understanding of the interplay of the substrate and residues within the active site of Sirtuin~5. Subsequently, we set out to compute the free energy as a function of the reaction coordinate instead of a minimum energy path. Another theme of this thesis is the computation of spectroscopic observables in a cost effective manner while simultaneously including important features of the experimental setup. The inclusion of solvent molecules and finite temperature effects has a decisive effect on the accuracy of the computed observables. In this context, we highlight the importance of sampling atomic configurations (with and without explicit solvent) and the non-negligible influence of electron correlation on the accuracy of computed observables. Simulation protocols are developed that enable sampling, the inclusion of correlation methods, and large quantum mechanical subsystems at a low computational cost.
Not available
Dietschreit, Johannes
2020
Englisch
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Dietschreit, Johannes (2020): Advanced techniques for the computer simulation and analysis of biomolecular systems. Dissertation, LMU München: Fakultät für Chemie und Pharmazie
[thumbnail of Dietschreit_Johannes.pdf]
Vorschau
PDF
Dietschreit_Johannes.pdf

33MB

Abstract

The Helmholtz free energy is one of the central quantities of classical thermodynamics, as it governs important chemical properties such as equilibria or reaction kinetics. It is, therefore, a desirable quantity to measure, predict, and understand. Unsurprisingly, many methods exist to compute free energy differences between two states of a system. In this thesis, the density of states integration method (DSI) is developed; it detects which subsystems mainly contribute to the free energy difference. The method utilizes the velocity density of states function (VDoS) of each atom to calculate its contribution to the vibrational free energy. It is possible without any approximation to assign fractions of the vibrational free energy to meaningful subsystems, where the local free energy difference is the sum over all atoms comprising that subsystem. In this way, large local changes can be identified (free energy hot-spots), which is crucial for the understanding of free energy differences. The validity and usefulness of DSI is shown via several examples and comparison with state of the art free energy methods. In addition to the development of DSI, this thesis also focuses on free energy barriers in the context of investigating the reaction mechanism of Sirtuin~5, a lysine deacylase class~III. The relationship between the configuration of the enzyme's active site and the height of the reaction barrier is studied by computing minimal energy paths for the catalyzed reaction starting from many different (educt) configurations. Using the power of machine learning, atom-atom distances influencing the activation barrier are identified, allowing for a comprehensive understanding of the interplay of the substrate and residues within the active site of Sirtuin~5. Subsequently, we set out to compute the free energy as a function of the reaction coordinate instead of a minimum energy path. Another theme of this thesis is the computation of spectroscopic observables in a cost effective manner while simultaneously including important features of the experimental setup. The inclusion of solvent molecules and finite temperature effects has a decisive effect on the accuracy of the computed observables. In this context, we highlight the importance of sampling atomic configurations (with and without explicit solvent) and the non-negligible influence of electron correlation on the accuracy of computed observables. Simulation protocols are developed that enable sampling, the inclusion of correlation methods, and large quantum mechanical subsystems at a low computational cost.