Bioisosterism is a field that is widely applied to biological molecules, including drugs and agrochemicals. Bioisosterism is the replacement of an active fragment in a molecule with another fragment similar in activity. The replacement is designed to alter the behavior of the molecule in its target environment. In previous work a bioisostere database called the Quantum Isostere Database (QID) was built out of descriptors derived from the theory of Quantum Chemical Topology (QCT). The current work aims to expand the existing QID to include ring fragments. A series of rings were characterised by QCT properties taken from the ring. It was found that four features of a ring each independently have a systematic effect on the ring’s properties. In other words, each of the characteristics of a ring can be changed and have the same effect on the ring’s properties irrespective of the other ring features. The rings were also characterised using the three QCT properties taken from a point within the ring. The three properties established a space where rings were positioned based on their respective three properties. The positions of the rings showed that the space was able to discern between ring types, and that the features of a ring could be predicted if only its three properties were known. To improve the QID the alignment method and scoring were tested. The alignment procedure is unable to correctly align collinear fragments. Therefore, a principal axis alignment procedure was successfully employed to align collinear fragments. For terminal fragments an alternative alignment procedure was proposed to account for the increased rotational freedom. A global axis system meant that the direction dependent properties for all fragments were expressed in this new axis system. This idea was extended further and it was found that the geometry of a molecule was imprinted in the electrostatics when they were expressed in the global axis system. Finally, a pKa prediction method which correlates a single ab initio bond length was tested against two data sets (enols and guanidines). The method relies on subsets to form, where molecules within a subset share a chemical or structural commonality. These subsets were able to distinguish between the five tautomeric forms for the guanidines and different conformations for the enols. All predictions were within 1.0 pKa units of experimental values.

<p>The prediction and characterisation of chemical and biochemical processes are fundamental tasks in computational chemistry. Small chemical systems can be characterised by the stationary points on potential energy surface and reaction paths linking them. For large biological systems, statistical sampling is required to characterising their average properties.</p> <p>This thesis presents my Ph.D. work on developing new methods to predict and characterise chemical and biological processes. Two path-finding methods for finding the minimum energy reaction path and alternative reaction paths for small gas-phase reactions have been elucidated with examples, and molecular dynamic simulations have been used to characterise the binding affinity of protein-ligand complex and the free energy of protonation processes in a protein.</p> <p>Specifically, the fast marching method (FMM) has been used to find the minimum energy path (MEP) on the potential energy surface (PES) for small gas-phase reactions. In this thesis, FMM is shown to be one of the most general and reliable surface-walking algorithms for finding the MEP. However, it is an expensive method. Some improvements have been illustrated in chapter 2 and chapter 3.</p> <p>I also proposed a new method (called QSM-NT) for finding all stationary points, accordingly all alternative reaction paths on the PES. Unlike other path-finding methods, QSM-NT overcomes the need of an initial guess of the path, and it can find all stationary points on the PES. QSM-NT has been proven to be efficient and reliable through applications on analytical PES and real chemical reaction. The difficulties and pitfalls associated with QSM-NT have been elucidated with examples.</p> <p>Molecular dynamic (MD) simulation and associated postprocessing procedures have been used to study the binding properties of caffeine-A_2A complex. The binding affinities of different binding modes have been calculated using MM/PBSA method. The binding pocket has been characterised with MM/GBSA energy decomposition. Our computational work provides significant insight to the targeted drug design of the adenosine A_2A receptor.</p> <p>The pH-dependent properties of a protein play important roles in the fundamental biological processes. The protonation states, namely, the pK_a values of ionisable residues, especially active-site residues are the prerequisites to understanding of the mechanisms of many biological processes. In this thesis, acetoacetate decarboxylase (AADase) is used as a test case for studying different types of pK_a prediction methods. Our computational results have shown that the site-site interactions from other ionisable residues are crucial to the pK_a prediction of the target residue.</p> <p>This thesis covers the range from small gas phase reaction prediction to large complex biological systems characterisation using quantum mechanical and molecular mechanical methods.</p> / Doctor of Philosophy (PhD)

<p>Understanding and predicting chemical phenomena is the main goal of computational chemistry. In this thesis I present my work on applying computational approaches to study chemical processes in biological and supramolecular systems.</p> <p>pH-responsive molecular tweezers have been proposed as an approach for targeting drug-delivery to tumors, which tend to have a lower pH than normal cells. In chapter 2 I present a computational study I performed on a pH-responsive molecular tweezer using <em>ab initio</em> quantum chemistry in the gas phase and molecular dynamics simulations in solution. The binding free energy in solution was calculated using Steered Molecular Dynamics. We observe, in atomistic detail, the pH-induced conformational switch of the tweezer and the resulting release of the drug molecule. Even when the tweezer opens, the drug molecule remains near a hydrophobic arm of the molecular tweezer. Drug release cannot occur, it seems, unless the tweezer is a hydrophobic environment with low pH.</p> <p>The protonation state of amino acid residues in proteins depends on their respective pK_a values. Computational methods are particularly important for estimating the pK_a values of buried and active site residues, where experimental data is scarce. In chapter 3 I used the cluster model approach to predict the pK_a of some challenging protein residues and for which methods based on the numerical solution of the Poisson-Boltzmann equation and empirical approaches fail. The ionizable residue and its close environment were treated quantum mechanically, while the rest of the protein was replaced by a uniform dielectric continuum. The approach was found to overestimate the electrostatic interaction leading to predicting lower pK_a values.</p> / Master of Science (MSc)