One of the challenges at the forefront of computational chemistry is the development of methods to simulate, and therefore make predictions about, biomolecular systems. A biomolecular system is any chemical system that is biologically relevant. These systems are often key to our understanding of life and encompass things such as proteins, enzymes and interactions with drug molecules to name but a few. The field of biomolecular simulation has a long history and is populated with a multitude of mature methods. Force fields are perhaps the most widely used tool to simulate these systems at large length and time scales. While force fields have historically enjoyed some success, it has been known since their inception that the approximations inherent in the methods are potentially limiting. Important effects such as polarisation and many-body effects have, until more recently, been ignored because of the high computational cost of including them. In recent decades a new state-of-the-art has emerged and force fields are increasingly accounting for these more complex effects as well as drawing from quantum mechanics. This thesis details the ongoing development of one such novel, state-of-the-art force field: FFLUX. Drawing on the theory of quantum chemical topology and leveraging modern machine learning techniques, FFLUX is unique in its methodology. Much of the work presented here concerns the development of theory and code crucial for FFLUX. For example, a modified version of the smooth particle mesh Ewald sum is derived that includes extra force terms arising from explicit dependence of multipole moments on atomic position. The result of this development is a working force field that is for the first time capable of simulating bulk biomolecular systems on nanosecond timescales. Alongside the development work, the very first application of FFLUX to bulk simulation is presented here. FFLUX is used to simulate liquid water, an important biological solvent. The predicted bulk properties are validated against experiment, the most important test of any theoretical method, as well as compared to other force fields. FFLUX stands up to this test and is shown to be a viable contender in the space of state-of-the-art force fields.
|Date of Award
|31 Dec 2022
- The University of Manchester
|Paul Popelier (Supervisor) & Samuel De Visser (Supervisor)