Search

Home > Research > Computation & Modelling Lab
Research

Computation & Modelling Laboratory

Parallel Computing

A major activity at SCRU is the porting of popular scientific software used in chemistry, biophysics and materials science to Graphical Processing Units (GPU). The SCRU laboratories have developed a Quantum Supercharger Library capable of accelerating legacy codes such as GAMESS-UK, GAMESS-US and NWChem on hardware accelerators such as GPGPUs. Kevin Naidoo who directs SCRU research has been collaborating with Quantum code developers, Paul Sherwood and Martyn Guest in the UK to accelerate the GAMESS-UK code on a GPU cluster.

References:

Kyle D. Fernandes, C. Alicia Renison and Kevin J. Naidoo. Quantum Supercharger Library: Hyper-Parallelism of the Hartree Fock Method. J. Comp. Chem. 2015, 36, 1399-1409.

C. Alicia Renison, Kyle D. Fernandes and Kevin J. Naidoo. Quantum Supercharger Library: Hyper-Parallel Integral Derivatives Algorithms for Ab Initio QM/MM Dynamics. J. Comp. Chem. 2015, 36, 1410-1419

Wilkinson, K.; Sherwood, P.; Guest, M. F.; Naidoo, K. J., Acceleration of the GAMESS-UK Electronic Structure Package on Graphical Processing Units. J. Comp. Chem. 2011, 32, 2313-2318.

Kevin J. Naidoo*, Karl Wilkinson and Kyle Fernandes, Accelerating Scientific Computing Code in FORTRAN: The Quantum Chemistry Project; PGI News October 2011 (http://www.pgroup.com/lit/articles/insider/v3n3a1.htm)

QM/MM Computations

We develop methods for chemical biological applications. Our laboratory designed a new method to cut polar (glycosidic bonds) making it possible to accurately compute complex glycans using a mixed quantum classical model. We call this the Simple Link Atom Hybrid Saccharide (SLASH) method. We have developed semi-empirical quantum methods to model chemical glycobiological events that we termed AM1/d-CB1. We are able to simulate glycosylation reactions more accurately using these methods that are specifically designed for chemical glycobiological applications.

References:

Werner Crous, Martin J. Field, and Kevin J. Naidoo, Simple Link Atom Saccharide Hybrid (SLASH) Treatment for Glycosidic Bonds at the QM/MM Boundary, J. Chem. Theor. Comput. 2014 10 (4), 1727-1738

Krishna Govender, Jiali Gao and Kevin J. Naidoo. AM1/d-CB1: A Semiempirical Model for QM/MM Simulations of Chemical Glycobiology Systems. J. Chem. Theor. Comput. 2014, 10, 4694-4707

Barnett, C. B.; Naidoo, K. J., Ring puckering: A metric for evaluating the accuracy of AM1, PM3, PM3CARB-1 and SCC-DFTB carbohydrate QM/MM simulations. J. Phys. Chem. B. 2010, 114, 17142-17154

Free Energy Computations

We take the view that the Free Energy of a system is a highly informative means with which to study it and design and improve on it. To this end have developed a method capable of calculating the Free Energy from Adaptive Reaction Coordinate Forces (FEARCF) for molecular conformation, association, and chemical reactions{Rajamani, 2003 #1666} in multiple dimensions.

References:

Naidoo, K. J.; Brady, J. W., Calculation of the Ramachandran Potential of Mean Force for a Disaccharide in Aqueous Solution. J. Am. Chem. Soc. 1999, 121, 2244-2252.

Rajamani, R.; Naidoo, K. J.; Gao, J., Implementation of an Adaptive Umbrella Sampling Method for the Calculation of Multidimensional Potential of Mean Force of Chemical Reactions in Solution. J. Comp. Chem. 2003, 24, 1775–1781.

Naidoo, K. J., Multidimensional free energy volumes offer unique insights into reaction mechanisms, molecular conformation and association. Phys. Chem. Chem. Phys. 2012, 14, 9026-9036.

Strümpfer, J.; Naidoo, K. J., Computing free energy hypersurfaces for anisotropic intermolecular associations. J. Comp. Chem. 2010, 31, 308-316.

Coarse Grain Computations

Now that the human genome has been decoded the next step is to map out the complete set of protein structures that are translated from the genome. This is the field of proteomics. The major problem is to quickly solve 3D protein structures this is largely a computational problem. We develop tools to use a coarse grained potential functions embedded in a molecular dynamics calculations to fold and refine protein and glycan conformations. Our specific structural interests are glycosyltransferases, glycosidases and glucokinases.

References:

Naidoo, K. J.; Schnitker, J.; Weeks, J. D., Two-Dimensional Melting Revisited: Molecular Dynamics Simulations Initiated with Optical Microscopy Data., Mol. Phys. 1993, 80, 1-43.

Naidoo, K. J.; Schnitker, J., Melting of two-dimensional colloidal crystals: A simulation study of the Yukawa system. J. Chem. Phys. 1994, 100, 3114-3121.

Molecular Modelling

We apply the computational methods that we develop along with standard molecular dynamics and hybrid quantum classical simulation techniques to two classes of molecules. Firstly, we discover glycosidase reaction mechanisms, from which we develop protocols for rational drug design from carbohydrate scaffolds. A second class of molecules are those of solvents making up the condensed phase. Here we investigate solution properties and design Ionic Liquid properties.

References

Matthews, R. P.; Venter, G. A.; Naidoo, K. J., Using Solvent Binding and Dielectric Friction to Interpret the Hydration Behavior of Complex Anions. J. Phys. Chem. B 2011, 115, 1045–1055.

Boscaino, A.; Naidoo, K. J., The Extent of Conformational Rigidity Determines Hydration in Non Aromatic HexaCyclic Systems. J. Phys. Chem. B 2011, 115, 2608–2616.

Barnett, C. B.; Wilkinson, K. A.; Naidoo, K. J., Molecular Details from Computational Reaction Dynamics for the Cellobiohydrolase I Glycosylation Reaction. J. Am. Chem. Soc. 2011, 133, 19474-19482.

 

Ionic Liquid Property Prediction and Design

The most obvious definition of an ionic liquid is a substance in the liquid state, consisting of positive cations and negative anions. Left unmodified, this definition would include simple ionic mixtures such as liquid NaCl as well, as long as it is applied above its melting point of >800°C. Room temperature ionic liquids (RTILs), on the other hand, are ionic mixtures with melting points <100°C. Popular RTILs consist of organic cations such as alkyl substituted imidazoliums, pyridiniums, pyrrolidiniums or ammoniums combined with inorganic or organic cations such as the halides, tetrafluoroborate, triflate, bistriflimide and acetate.

ionic-01ionic-02

Ion pairs of ionic liquids 1–ethyl–3–methylimidazolium tetrafluoroborate (left) and 1–ethyl–3–methylimidazolium acetate

Depending on the chosen combination of cation and anion, physical properties such as melting point, viscosity and conductivity vary significantly. Within a family, further changes such as the length of the alkyl chain or functionalisation with e.g. ether groups, can also lead to a host of varying properties.

With a liquidus range of several hundreds of degrees (often the upper temperature limit is not governed by vaporization, but by decomposition) and a wide range of physical properties, the RTILs have numerous exciting applications. With the promise of being able to engineer an ionic liquid that suits the requirements of a specific application, the RTILs make ideal candidates to replace traditional molecular solvents in organic reactions, metal catalysis and electrochemistry.

Simulation and Calculation of RTIL Properties

Whichever application is focused on, a fundamental requirement common to all is a thorough understanding of the nature of the physicochemical properties of the RTILs. The idea of compiling structure-property relationships is intriguing, but rests on a solid understanding and prediction of the intermolecular interactions and related dynamic behaviour of these systems, and it is here where microscopic insight is needed that molecular simulation and calculation becomes a powerful tool. Quantum mechanical (QM) calculations can be used to investigate the electronic stucture of ions and ion pairs. Correctly chosen methods provide accurate descriptions of ion–ion interaction, polarization, hydrogen bonding and dispersion interaction. Molecular dynamics (MD) simulations can be used to integrate over the time evolution of a system, subject to a complete (and accurate) formulation of the intermolecular forces, which can subsequently provide further thermodynamic information and transport properties.

ionic-03

Hydrogen bonding in a 1–ethyl–3–methylimidazolium chloride ion pair as identified through critical points in the electron density gradient (QTAIM) and the noncovalent index (NCI).


Developing structure–property relationships for room temperature ionic liquids require computational methods (static and polarizable force fields) that accurately describe the range of intermolecular interactions present. Developing such computational methods is one of the aims of this research. Furthermore, important insight into structure–property relationships can be gained through a better understanding of the nature of the interactions, and the effect of strengthening/weakening particular components of it.