Author ORCID Identifier


Defense Date


Document Type


Degree Name

Doctor of Philosophy


Pharmaceutical Sciences

First Advisor

Glen Kellogg


Traditional drug discovery has rapidly accelerated thanks to development of computational molecular modeling. The crucial component that these computational studies hinge upon is having a well-defined, and energetically favorable structure. Structures of proteins and ligands that meet these criteria are important for accurately simulating models used to study drug binding. To demonstrate the role of accurate structure simulation in the study of these events, this thesis presents, first, a story examining the problem of accurate structure modeling of ionizable residues within protein structures, specifically aspartic acid, glutamic acid, and histidine. I present our method, which uses the HINT force field to simulate “titration” of these residues and study which hydropathic environments may contribute to stabilization of certain protonation states. We further use this data to construct and cluster together pH-tunable hydropathic interaction maps, detailing the kinds of interactions these residues make with their environments in low and high-pH situations. xxii The second story describes identification of new, potent inhibitors against eIF4A1 (eukaryotic initiation factor 4A1), driven by computational techniques. This story describes a pharmacophoric virtual screen of chemical databases for novel inhibitors, based on the structure of Rocaglamide A (RocA), a natural product inhibitor of eIF4A1. After docking and HINT scoring studies of hit compounds, we identified many highly potent compounds. Computational studies have yielded a reasonable binding mode for this series of compounds and suggest design of new, more potent compounds with better drug-like properties. The final story builds upon our compilation of hydropathic interaction maps in the design of a protein-protein interface optimization program that will be the roots of a protein-protein docking tool. We compile vast amounts of hydropathic map data, detailing what we call residue “hydropathic valences,” for this purpose. The tool implements a genetic algorithm for population-weighted choice of map combinations for residues at a protein-protein interface. Our model is currently being trained on publicly available, high-resolution crystal structures. We hope for development of this tool to be the beginning of returns made on a long series of chapters of data collection for this purpose. This thesis is a record of diligent efforts to apply HINT to novel drug discovery and protein structure prediction tools. It will demonstrate the integral role of using or creating accurate structure models for studying protein structure and how these studies may ultimately be used for development of new clinical therapeutics. Let this work also stand as a testament to the power of computational techniques to efficiently simulate real-world biomolecular events on an atomic scale in a way that even allows this translation from in silico theory to potentially in vivo reality. Let it be astounding to the reader, as it was for me.


© Noah Benjamin Herrington

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission