Document Type
Dissertation - Open Access
Award Date
2025
Degree Name
Doctor of Philosophy (PhD)
Department / School
Chemistry and Biochemistry
First Advisor
Suvobrata Chakaravarty
Abstract
Proteins are central to nearly all cellular processes, and their structural complexity underpins their functional diversity. Beyond covalent bonds and defined structural motifs, subtle weak non-covalent interactions—including anion–π, cation–π, CH–π, and salt bridges—govern protein folding, stability, dynamics, and interactions. Despite their importance, these forces remain poorly understood, especially in proteins lacking experimentally resolved structures. A deeper understanding of such interactions is essential not only for mechanistic insights into protein function but also for therapeutic targeting of protein–protein interactions (PPIs). This thesis bridges these gaps through computational tool development, integrative structural prediction, and collaborative experimental approaches. Chapter 2 introduces AQcalc, a Python/Django-based webserver that identifies weak non-covalent interactions across diverse protein classes, including membrane and post-translationally modified proteins. AQcalc also enables evaluation of AlphaFold predictions and was published in Protein Science (2023), where I served as first author. Chapter 3 presents a collaborative study on coevolving residues in PHD finger domains, integrating proteolysis, thermal unfolding, and generative model–based energy estimation to dissect residue-level contributions to stability. My analyses helped distinguish residues critical for folding from those with limited roles. Chapter 4 expands this focus to protein–protein interactions using SPOC (Statistical Potential of Complexes), a complementary tool that evaluates AlphaFold-Multimer predictions. SPOC distinguishes true from false interaction interfaces, correlates with binding affinities, and identifies biologically relevant complexes while filtering out decoys. Integrating SPOC with fragment-based modeling and proteomic datasets refines large-scale AI-generated predictions into testable hypotheses. Collectively, this thesis introduces new computational strategies for probing weak non-covalent interactions and their roles in protein stability and binding. By uniting AQcalc, experimental–computational analyses, and interface evaluation via SPOC, this work advances atomistic understanding of weak interactions and highlights translational opportunities in precision medicine and drug discovery.
Publisher
South Dakota State University
Recommended Citation
Afshinpour, Maral, "Integrative Computational Approaches to Protein Structure, Dynamics, and Function: Insights Into Weak Interactions and Their Biological Implications" (2025). Electronic Theses and Dissertations. 1881.
https://openprairie.sdstate.edu/etd2/1881
Included in
Rights Statement
