Novel Enhanced Sampling Methods in Multiscale Modeling
Lead Research Organisation:
University College London
Department Name: Physics and Astronomy
Abstract
Computer-based technologies are becoming one of the most promising novel approaches due to continuously accelerated growth of both hardware processing power and software algorithm efficiency. One recent example includes machine learning algorithms that revolutionised data analysis in computer science, and lead to new computer games, visual recognition, and other applications that overtake human performance in many cases.
Here, we propose to perform atomistic molecular simulations using novel enhanced sampling algorithms. Most biologically important processes take place on significantly longer timescales than those accessible to current computer simulations. Therefore, to obtain meaningful and accurate results regarding the kinetics and conformational dynamics of complex molecular systems, we use algorithms that enhance the sampling using parallel calculations with different biases. Developing more optimal biasing algorithms will allow us to model faster and more accurately the key biological processes of interest, including ligand binding, protein conformations, etc.
Here we aim to use statistical algorithms inspired by machine learning to develop novel enhanced sampling methods for molecular simulations. Novel algorithms can be applied to a wide range of molecular modeling problems. We will focus on phosphate catalytic enzymes, and study key DNA processing enzymes to reveal the catalytic mechanism in these systems.
Due to the essential nature of phosphate catalytic enzymes in most biological processes, a large number of drugs in current clinical practice also target phosphate-processing enzymes treating a wide range of diseases. Examples include reverse transcriptase and integrase inhibitors used against HIV and hepatitis B, proton pump inhibitors used in gastric diseases, kinase, PARP and topoisomerase inhibitors used against a large number of cancers. Studying phosphate catalytic systems with modern molecular modeling methods will enable fundamental advances in our current knowledge of the molecular basis of life. It will also create opportunities for rational development of better drugs to fight diseases.
Here, we propose to perform atomistic molecular simulations using novel enhanced sampling algorithms. Most biologically important processes take place on significantly longer timescales than those accessible to current computer simulations. Therefore, to obtain meaningful and accurate results regarding the kinetics and conformational dynamics of complex molecular systems, we use algorithms that enhance the sampling using parallel calculations with different biases. Developing more optimal biasing algorithms will allow us to model faster and more accurately the key biological processes of interest, including ligand binding, protein conformations, etc.
Here we aim to use statistical algorithms inspired by machine learning to develop novel enhanced sampling methods for molecular simulations. Novel algorithms can be applied to a wide range of molecular modeling problems. We will focus on phosphate catalytic enzymes, and study key DNA processing enzymes to reveal the catalytic mechanism in these systems.
Due to the essential nature of phosphate catalytic enzymes in most biological processes, a large number of drugs in current clinical practice also target phosphate-processing enzymes treating a wide range of diseases. Examples include reverse transcriptase and integrase inhibitors used against HIV and hepatitis B, proton pump inhibitors used in gastric diseases, kinase, PARP and topoisomerase inhibitors used against a large number of cancers. Studying phosphate catalytic systems with modern molecular modeling methods will enable fundamental advances in our current knowledge of the molecular basis of life. It will also create opportunities for rational development of better drugs to fight diseases.
Planned Impact
Academia
Our results are shared via workshops, tutorials, conferences and seminars. Computational groups will be able to download our newly developed programs from my research group website (www.rostaresearch.com), or from CCPBioSim workshop/tutorial webpages (http://www.ccpbiosim.ac.uk/). Our experimental collaborators will benefit from the more accurate and efficient algorithms that we can subsequently apply to design novel ligands, mutants, and test experimental hypothesis. We will work closely with experimental and theoretical collaborator groups in the UK, and overseas.
Public Sector, Business, Industry
On long term, health-related public sectors will benefit from basic research on structure and mechanism of phosphate processing enzymes. Our methods can be used and may be inspirational to a large number of projects studying ligand binding kinetics, or the dynamics of phosphate-processing enzymes that are relevant to many diseases. Phosphate processing enzymes are validated targets of a large number of drugs used in current clinical practices treating a wide range of diseases. These include reverse transcriptase and integrase inhibitors used against HIV and hepatitis B, proton pump inhibitors used in gastric diseases, kinase and topoisomerase inhibitors used in chemotherapy to treat cancers.
Our basic research results are therefore also relevant to UK charities such as Cancer Research UK. In addition to drug design, insights related to controlling enzyme activity is also relevant for biotechnology industries, e.g., businesses developing industrial enzymes such as Novozymes.
General Public, Education
The general public, high school and university students will benefit from new basic research developments in general, by public lectures in the UK and world-wide (e.g., via STEM seminars, such as the one I presented in a London-area Girls' School), or by the Open Days at King's. My lab also hosted 10 high school students to date since 2013, who were introduced to on-going research in my lab via the In2Science and Nuffield Research Placement programs.
Our results are shared via workshops, tutorials, conferences and seminars. Computational groups will be able to download our newly developed programs from my research group website (www.rostaresearch.com), or from CCPBioSim workshop/tutorial webpages (http://www.ccpbiosim.ac.uk/). Our experimental collaborators will benefit from the more accurate and efficient algorithms that we can subsequently apply to design novel ligands, mutants, and test experimental hypothesis. We will work closely with experimental and theoretical collaborator groups in the UK, and overseas.
Public Sector, Business, Industry
On long term, health-related public sectors will benefit from basic research on structure and mechanism of phosphate processing enzymes. Our methods can be used and may be inspirational to a large number of projects studying ligand binding kinetics, or the dynamics of phosphate-processing enzymes that are relevant to many diseases. Phosphate processing enzymes are validated targets of a large number of drugs used in current clinical practices treating a wide range of diseases. These include reverse transcriptase and integrase inhibitors used against HIV and hepatitis B, proton pump inhibitors used in gastric diseases, kinase and topoisomerase inhibitors used in chemotherapy to treat cancers.
Our basic research results are therefore also relevant to UK charities such as Cancer Research UK. In addition to drug design, insights related to controlling enzyme activity is also relevant for biotechnology industries, e.g., businesses developing industrial enzymes such as Novozymes.
General Public, Education
The general public, high school and university students will benefit from new basic research developments in general, by public lectures in the UK and world-wide (e.g., via STEM seminars, such as the one I presented in a London-area Girls' School), or by the Open Days at King's. My lab also hosted 10 high school students to date since 2013, who were introduced to on-going research in my lab via the In2Science and Nuffield Research Placement programs.
Publications
Koczor-Benda Z
(2022)
Molecular Vibration Explorer: an Online Database and Toolbox for Surface-Enhanced Frequency Conversion and Infrared and Raman Spectroscopy.
in The journal of physical chemistry. A
Hu S
(2023)
Full Control of Plasmonic Nanocavities Using Gold Decahedra-on-Mirror Constructs with Monodisperse Facets.
in Advanced science (Weinheim, Baden-Wurttemberg, Germany)
Griffiths J
(2021)
Resolving sub-angstrom ambient motion through reconstruction from vibrational spectra.
in Nature communications
Dürr S
(2021)
The Role of Conserved Residues in the DEDDh Motif: the Proton-Transfer Mechanism of HIV-1 RNase H
in ACS Catalysis
Berta D
(2021)
Modelling the active SARS-CoV-2 helicase complex as a basis for structure-based inhibitor design.
in Chemical science
Badaoui M
(2022)
Combined Free-Energy Calculation and Machine Learning Methods for Understanding Ligand Unbinding Kinetics
in Journal of Chemical Theory and Computation
| Description | Large kinetic networks are central to many data driven disciplines, yet clustering is often done by disregarding the dynamical information of the system. We developed a novel clustering algorithm to enable dynamical coarse graining of complex kinetic networks that can reduce the size of the network, making it more tractable, while keeping the key kinetic properties of the network preserved. Unbinding rates have important role in determining ligand efficiencies, and can be more relevant than binding constants, however, they are very difficult to determine both in experiments and via computational modelling. We developed novel methods to assess the unbinding paths and quantitative rates of the unbinding of small molecules from proteins, a work carried out in collaboration with Novartis scientists. |
| Exploitation Route | Our methods can provide important insight for determining drug efficiency by providing atomic-level details of the complex unbinding processes. |
| Sectors | Pharmaceuticals and Medical Biotechnology |