On using automated algorithms to parameterise molecules for molecular dynamics simulations and investigating suitable ensembles for the simulation of naphthalimide monolayers : a thesis submitted to Massey University in Albany, Auckland in fulfilment of the requirements for the degree of Doctor of Philosophy in Chemistry, Massey University, Auckland
Molecular dynamics simulations provide a means to investigate the spatial and temporal evolution
of systems of molecules at atomic resolution. Force fields are used to describe the interactions
between atoms contained within the system. A number of such force fields have been
developed over the years, with a focus on force fields for use in simulations of biochemical systems,
in particular, protein systems. This thesis is primarily focused on extending the range of
systems that can be simulated through providing means for automated generation of force field
parameters for large novel molecules.
One component of existing force fields that is generally poorly parameterised are the dihedral
terms. In combination with the non-bonded terms, the dihedral terms are used to describe the rotational
energy profile about bonds, and have a large influence on the conformational properties
of a simulated system. A new method for the determination of dihedral parameters is developed,
utilising high level quantum mechanical calculations. With the use of local elevation molecular
dynamics simulations, this method is applied to the case of protein backbone dihedrals within
the GROMOS force field.
When one desires to simulate the interaction of a novel molecule with some biochemical system,
the novel molecule must be parameterised in a manner that is compatible with the force
field used to describe the biochemical system. However, doing so is a slow, tedious, and error
prone process, especially when the novel molecule is large. To combat this, a new algorithm,
known as CherryPicker, was developed. CherryPicker is a graph based algorithm which enables
rapid parameterisation of large molecules through fragment comparison with a library of previously
parameterised small molecules. The algorithm design is discussed and tested on a few
simple test cases in part II.
Part III steps away from the parameterisation focus of this thesis and looks at the simulation of
naphthalimide monolayers. Naphthalimides have applications in sensing environments as they
have absorption and fluorescence emission spectra lying within the UV and visible regions of
light. With a long chain alkane substituted at the N-imide site, they become amphiphilic and can
form monolayers on the surface of water, and can be transferred to a solid substrate when at a
desired compression level. Molecular dynamics simulations can be used to provide insight into
the formation of compressed monolayer phase. Here, the effect of different ensembles, namely
NVT, NPT, and NgT are investigated for use in simulating a naphthalimide monolayer.