Document Type


Date of Degree

Fall 2018

Degree Name

PhD (Doctor of Philosophy)

Degree In


First Advisor

Elcock, Adrian H

First Committee Member

Pufall, Miles A

Second Committee Member

Fuentes, Ernesto J

Third Committee Member

Schnieders, Michael J

Fourth Committee Member

Spies, M Ashley

Fifth Committee Member

Margulis, Claudio J


Computer simulations allow researchers to study the dynamics and interactions of biological molecules in ways that cannot be currently achieved in experiments. In this work, I have used computer simulations to study the following systems: (1) carbohydrate-carbohydrate and carbohydrate-amino acid interactions using all-atom molecular dynamics simulations, and (2) protein-protein interactions using coarse-grained implicit solvent models.

My first studies involved simulating carbohydrate and amino acid systems using atomistic force fields. During my initial simulations, I observed that carbohydrates were interacting too favorably leading them to aggregate in conditions under which they experimentally remain soluble. To alleviate this issue, I surgically modified the carbohydrate-carbohydrate interaction parameters in order to match osmotic pressure data from experiment. This approach was successful while preserving many of the correct features of the original force field. Next, I observed similar issues in carbohydrate-amino acid simulations and used the same methodology to correct carbohydrate-amino acid parameters. I showed that the modified parameters also worked well in simulations of much larger systems, allowing realistic simulations to be performed on polymeric sugars such as dextran and the peptidoglycan layer of the cell wall.

In a more recent and separate study, I have attempted to parameterize very coarse-grained models of proteins (for eventual use in cellular scale simulations) using experimental osmotic second virial coefficients. I found that a 10 residue-per-bead model including electrostatic interactions could approximately match most of the second virial coefficient data obtained from experiment. In contrast, a more simplified, spherical model of proteins could not adequately reproduce experiment. Although more work will be required to establish a better quantitative agreement with experiment, my results indicate that even very coarse models of proteins can produce reasonably accurate simulations of protein-protein interactions.


Computer Simulations, Molecular Dynamics


xiii, 135 pages


Includes bibliographical references (pages 122-135).


Copyright © 2018 Wesley K. Lay

Included in

Biochemistry Commons