| 16 |
|
|
| 17 |
|
\frontmatter |
| 18 |
|
|
| 19 |
< |
\title{APPLICATION AND DEVELOPMENT OF MOLECULAR DYNAMICS TECHNIQUES FOR THE |
| 19 |
> |
\title{DEVELOPMENT OF MOLECULAR DYNAMICS TECHNIQUES FOR THE |
| 20 |
|
STUDY OF WATER AND OTHER BIOCHEMICAL SYSTEMS} |
| 21 |
|
\author{Christopher Joseph Fennell} |
| 22 |
|
\work{Dissertation} |
| 29 |
|
|
| 30 |
|
\begin{abstract} |
| 31 |
|
|
| 32 |
< |
The following dissertation lays out research that I have performed |
| 33 |
< |
over the last several years. All of the work relies on the technique of |
| 34 |
< |
molecular dynamics, and in this dissertation I start by outlining many |
| 35 |
< |
of the considerations that go into molecular dynamics |
| 36 |
< |
simulations. This is followed by an introduction to {\sc oopse}, the |
| 37 |
< |
object oriented parallel simulation engine, which is a program for |
| 38 |
< |
performing molecular simulations developed and maintained in our |
| 39 |
< |
lab. Most of the research was performed either using {\sc oopse} or |
| 40 |
< |
earlier code that predated {\sc oopse}. |
| 32 |
> |
This dissertation comprises a body of research in the field of |
| 33 |
> |
classical molecular simulations, with particular emphasis placed on |
| 34 |
> |
the proper depiction of water. This work is arranged such that the |
| 35 |
> |
techniques and models used within are first developed and tested |
| 36 |
> |
before being applied and compared with experimental results. With this |
| 37 |
> |
organization in mind, it is appropriate that the first chapter deals |
| 38 |
> |
primarily the technique of molecular dynamics and technical |
| 39 |
> |
considerations needed to correctly perform molecular simulations. |
| 40 |
|
|
| 41 |
< |
This introduction is followed by three chapters that discuss in detail |
| 42 |
< |
the primary research projects for which I am responsible. The first |
| 43 |
< |
project discusses my work on electrostatic interaction correction |
| 44 |
< |
techniques, with applications to water and biologically relevant |
| 45 |
< |
molecular systems. This leads into work on improving the depiction of |
| 46 |
< |
water in molecular simulations by refining simple and highly |
| 47 |
< |
computationally efficient single point water models. The final project |
| 48 |
< |
discussed in this body of research involves free energy calculations |
| 49 |
< |
of ice polymorphs, and includes investigations of a new ice polymorph |
| 50 |
< |
that we discovered while performing simulations involving the single |
| 51 |
< |
point water models. |
| 41 |
> |
Building on this framework, the second chapter discusses correction |
| 42 |
> |
techniques for handling the long-ranged electrostatic interactions |
| 43 |
> |
common in molecular simulations. Particular focus is placed on a |
| 44 |
> |
shifted-force ({\sc sf}) modification of the damped shifted Coulombic |
| 45 |
> |
summation method. In this work, {\sc sf} is shown to be nearly |
| 46 |
> |
equivalent to the more commonly utilized Ewald summation in |
| 47 |
> |
simulations of condensed phases. Since the {\sc sf} technique is |
| 48 |
> |
pairwise, it scales as $\mathcal{O}(N)$ and lacks periodicity |
| 49 |
> |
artifacts introduced through heavy reliance on the reciprocal-space |
| 50 |
> |
portion of the Ewald sum. The electrostatic damping technique used |
| 51 |
> |
with {\sc sf} is then extended beyond simple charge-charge |
| 52 |
> |
interactions to include point-multipoles. Optimal damping parameter |
| 53 |
> |
settings are also determined to ensure proper depiction of the |
| 54 |
> |
dielectric behavior of molecular systems. Presenting this technique |
| 55 |
> |
early enables its application in the systems discussed in the later |
| 56 |
> |
chapters and shows how it can improve the quality of various molecular |
| 57 |
> |
simulations. |
| 58 |
|
|
| 59 |
< |
I end this dissertation with some concluding remarks and |
| 60 |
< |
appendices. The conclusion simply sums up the previous sections and |
| 61 |
< |
comments on the research findings. The appendices include supporting |
| 62 |
< |
information and a more detailed look at systems that were treated in a |
| 63 |
< |
more general form in the earlier sections. |
| 59 |
> |
The third chapter applies the above techniques and focuses on water |
| 60 |
> |
model development, specifically the single-point soft sticky dipole |
| 61 |
> |
(SSD) model. In order to better depict water with SSD in computer |
| 62 |
> |
simulations, it needed to be reparametrized. This work results in the |
| 63 |
> |
development of SSD/RF and SSD/E, new variants of the SSD model |
| 64 |
> |
optimized for simulations with and without a reaction field |
| 65 |
> |
correction. These new single-point models are more efficient than the |
| 66 |
> |
common multi-point partial charge models and better capture the |
| 67 |
> |
dynamic properties of water. SSD/RF can be successfully used with |
| 68 |
> |
damped {\sc sf} through the multipolar extension of the technique |
| 69 |
> |
described in the previous chapter. Discussion on the development of |
| 70 |
> |
the two-point tetrahedrally restructured elongated dipole (TRED) water |
| 71 |
> |
model is also presented, and this model is optimized for use with the |
| 72 |
> |
damped {\sc sf} technique. Though there remain some algorithmic |
| 73 |
> |
complexities that need to be addressed (logic for neglecting |
| 74 |
> |
charge-quadrupole interactions between other TRED molecules) to use |
| 75 |
> |
this model in general simulations, it is approximately twice as |
| 76 |
> |
efficient as the commonly used three-point water models (i.e. TIP3P |
| 77 |
> |
and SPC/E). |
| 78 |
|
|
| 79 |
+ |
Continuing in the direction of model applications, the final chapter |
| 80 |
+ |
deals with a unique polymorph of ice that was discovered while |
| 81 |
+ |
performing water simulations with the fast simple water models |
| 82 |
+ |
discussed in the previous chapter. This form of ice, called |
| 83 |
+ |
``imaginary ice'' (Ice-$i$), has a low-density structure which is |
| 84 |
+ |
different from any known polymorph observed in either experiment or |
| 85 |
+ |
computer simulation studies. The free energy analysis discussed here |
| 86 |
+ |
shows that this structure is in fact the thermodynamically preferred |
| 87 |
+ |
form of ice for both the single-point and commonly used multi-point |
| 88 |
+ |
water models under the chosen simulation conditions. It is shown that |
| 89 |
+ |
inclusion of electrostatic corrections is necessary to obtain more |
| 90 |
+ |
realistic results; however, the free energies of the various |
| 91 |
+ |
polymorphs (both imaginary and real) in many of these models is shown |
| 92 |
+ |
to be so similar that choice of system properties, like the volume in |
| 93 |
+ |
$NVT$ simulations, can directly influence the ice polymorph expressed. |
| 94 |
+ |
|
| 95 |
|
\end{abstract} |
| 96 |
|
|
| 97 |
|
\begin{dedication} |
| 113 |
|
thank Christopher Harrison and Dr. Steven Corcelli for additional |
| 114 |
|
discussions and comments. Finally, I would like to thank my parents, |
| 115 |
|
Edward P.~Fennell and Rosalie M.~Fennell, for providing the |
| 116 |
< |
opprotunities and encouragement that allowed me to pursue my |
| 116 |
> |
opportunities and encouragement that allowed me to pursue my |
| 117 |
|
interests, and I would like to thank my wife, Kelley, for her |
| 118 |
< |
unwaivering support. |
| 118 |
> |
unwavering support. |
| 119 |
|
\end{acknowledge} |
| 120 |
|
|
| 121 |
|
\mainmatter |
