Header

UZH-Logo

Maintenance Infos

Quantum mechanical methods for structure-based drug design


Zhou, Ting. Quantum mechanical methods for structure-based drug design. 2010, University of Zurich, Faculty of Science.

Abstract

Quantum mechanical (QM) methods are becoming popular in computational drug design and development mainly because high accuracy is required to estimate (relative) binding affinities. For low- to medium-throughput in silico screening, (e.g., scoring and prioritizing a series of inhibitors sharing the same molecular scaffold) efficient approximations have been developed in the past decade, like linear scaling QM in which the computation time scales almost linearly with the number of basis functions. The first chapter of this thesis is a review of QM methods for drug design.
In the second chapter, LIECE, which is short for linear interaction energy model with continuum electrostatic solvation, is further improved by using a linear scaling semiempirical QM method. The new method is termed QMLIECE. Then QMLIECE is tested on three enzyme/inhibitor systems: the West Nile virus NS3 protease (a serine protease), the HIV-1 protease (an aspartic protease), and the human cyclindependent kinase 2. After that the necessity of QM method for predicting binding free energy is discussed.
In the third chapter, the QM probe method is suggested as a filter in high throughput virtual screening. To speed up the calculation, the QM probe method approximates the ATP-binding site of the tyrosine kinase erythropoietin producing human hepatocellular carcinoma receptor B4 (EphB4) by 6 types of small molecules (probes) with particular orientations and uses a semi-empirical QM Hamiltonian to calculate interaction energies between a compound and probes. These interaction energies are further used for filtering multimillion poses generated by high throughput docking. A single-digit micromolar inhibitor of EphB4 with a relatively good selectivity profile is identified upon experimental tests of only 23 molecules.
To elucidate the technology of high throughput docking used in the third chapter, in the forth chapter, a distributed virtual screening data management system (DVSDMS) is introduced, in which the data handling and the distribution of jobs are realized by the open-source structured query language database software MySQL. In DVSDMS, the data management is separated from docking and ranking process. In both benchmark and production, DVSDMS performed efficiently with a limited effort of programming and a trivial investment of software and hardware. In the last chapter, the original ultrafast shape recognition (USR) method is complemented with an optical isomerism descriptor (USR:OptIso). Then the USR:OptIso is tested on discriminating mirror images of three kinase inhibitors and 15 types of isomers, some of which can not be distinguished by the original USR, which uses only the atomic distances. Finally, both similarity scores calculated by the original USR and USR:OptIso are extensively compared with ROCS shape Tanimoto, which is based on Gaussian molecular volume overlap.


Quantenmechanische (QM) Methoden erfreuen sich im computergestützen Wirkstoffdesign und der Entwicklung wachsender Beliebtheit, da eine hohe Genauigkeit für die Berechnung von Bindungsaffinitäten notwendig ist. Für das Screening einer kleinen und mittleren Anzahl an Molekülen wurden im letzten Jahrzehnt effiziente Näherungen entwickelt. In dieser Arbeit wird LIECE, was kurz ist für lineares Interaktions-Energiemodell mit kontinuierlicher elektrostatischer Solvatation, weiterentwickelt durch Verwendung einer linear skalierenden, semiempirischen QM Methode. Das neue Verfahren wird QMLIECE genannt. QMLIECE wird an drei Enzym/Inhibitor-Komplexen getestet. Anschliessend wird die Notwendigkeit der QM Methode für die Vorhersage der freien Bindungsenergien diskutiert.
Im nächsten Abschnitt der Arbeit wird die QM Probemethode als Filter für virtuelles Hochdurchsatz-Screening vorgeschlagen, für das ein neues virtuelles Screening-Datenverarbeitungssystem angewendet wird. Ein einstellig mikromolarer Inhibitor von EphB4 mit relativ gutem Selektivitätsprofil wird identifiziert durch experimentelle Untersuchung von nur 23 Molekülen.

Abstract

Quantum mechanical (QM) methods are becoming popular in computational drug design and development mainly because high accuracy is required to estimate (relative) binding affinities. For low- to medium-throughput in silico screening, (e.g., scoring and prioritizing a series of inhibitors sharing the same molecular scaffold) efficient approximations have been developed in the past decade, like linear scaling QM in which the computation time scales almost linearly with the number of basis functions. The first chapter of this thesis is a review of QM methods for drug design.
In the second chapter, LIECE, which is short for linear interaction energy model with continuum electrostatic solvation, is further improved by using a linear scaling semiempirical QM method. The new method is termed QMLIECE. Then QMLIECE is tested on three enzyme/inhibitor systems: the West Nile virus NS3 protease (a serine protease), the HIV-1 protease (an aspartic protease), and the human cyclindependent kinase 2. After that the necessity of QM method for predicting binding free energy is discussed.
In the third chapter, the QM probe method is suggested as a filter in high throughput virtual screening. To speed up the calculation, the QM probe method approximates the ATP-binding site of the tyrosine kinase erythropoietin producing human hepatocellular carcinoma receptor B4 (EphB4) by 6 types of small molecules (probes) with particular orientations and uses a semi-empirical QM Hamiltonian to calculate interaction energies between a compound and probes. These interaction energies are further used for filtering multimillion poses generated by high throughput docking. A single-digit micromolar inhibitor of EphB4 with a relatively good selectivity profile is identified upon experimental tests of only 23 molecules.
To elucidate the technology of high throughput docking used in the third chapter, in the forth chapter, a distributed virtual screening data management system (DVSDMS) is introduced, in which the data handling and the distribution of jobs are realized by the open-source structured query language database software MySQL. In DVSDMS, the data management is separated from docking and ranking process. In both benchmark and production, DVSDMS performed efficiently with a limited effort of programming and a trivial investment of software and hardware. In the last chapter, the original ultrafast shape recognition (USR) method is complemented with an optical isomerism descriptor (USR:OptIso). Then the USR:OptIso is tested on discriminating mirror images of three kinase inhibitors and 15 types of isomers, some of which can not be distinguished by the original USR, which uses only the atomic distances. Finally, both similarity scores calculated by the original USR and USR:OptIso are extensively compared with ROCS shape Tanimoto, which is based on Gaussian molecular volume overlap.


Quantenmechanische (QM) Methoden erfreuen sich im computergestützen Wirkstoffdesign und der Entwicklung wachsender Beliebtheit, da eine hohe Genauigkeit für die Berechnung von Bindungsaffinitäten notwendig ist. Für das Screening einer kleinen und mittleren Anzahl an Molekülen wurden im letzten Jahrzehnt effiziente Näherungen entwickelt. In dieser Arbeit wird LIECE, was kurz ist für lineares Interaktions-Energiemodell mit kontinuierlicher elektrostatischer Solvatation, weiterentwickelt durch Verwendung einer linear skalierenden, semiempirischen QM Methode. Das neue Verfahren wird QMLIECE genannt. QMLIECE wird an drei Enzym/Inhibitor-Komplexen getestet. Anschliessend wird die Notwendigkeit der QM Methode für die Vorhersage der freien Bindungsenergien diskutiert.
Im nächsten Abschnitt der Arbeit wird die QM Probemethode als Filter für virtuelles Hochdurchsatz-Screening vorgeschlagen, für das ein neues virtuelles Screening-Datenverarbeitungssystem angewendet wird. Ein einstellig mikromolarer Inhibitor von EphB4 mit relativ gutem Selektivitätsprofil wird identifiziert durch experimentelle Untersuchung von nur 23 Molekülen.

Statistics

Downloads

686 downloads since deposited on 24 Jan 2011
69 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Dissertation (monographical)
Referees:Caflisch Amedeo, Baldridge Kim K
Communities & Collections:04 Faculty of Medicine > Department of Biochemistry
07 Faculty of Science > Department of Biochemistry

07 Faculty of Science > Department of Chemistry
UZH Dissertations
Dewey Decimal Classification:570 Life sciences; biology
Language:English
Place of Publication:Zürich
Date:2010
Deposited On:24 Jan 2011 16:45
Last Modified:19 May 2020 15:11
Number of Pages:144
Additional Information:Enthält Sonderdrucke
OA Status:Green
Related URLs:https://www.recherche-portal.ch/permalink/f/5u2s2l/ebi01_prod006250086 (Library Catalogue)

Download

Green Open Access

Download PDF  'Quantum mechanical methods for structure-based drug design'.
Preview
Content: Published Version
Language: English
Filetype: PDF
Size: 8MB