Back to search

Cornell University

Constraint-Based Methods for Macromolecular Structure Prediction and Multi-Conformer Refinement

Abstract

dc:description.abstract

This dissertation addresses fundamental challenges in structural biology through the development and application of constraint satisfaction formulations, which are effectively solved using iterative projection methods such as Reflect-Reflect-Relax (RRR). We demonstrate how the divide-and-concur framework, when combined with these methods, can successfully tackle complex structural problems that require the satisfaction of multiple constraints involving molecular geometry, X-ray contrast (electron charge density), and energy models. We begin with a proof-of-concept study involving simplified two-dimensional lattice proteins, composed of only two residues, H (hydrophobic) and P (polar), designed to capture the essential features of globular protein folding. This study establishes the viability of constraint-based approaches for structural prediction problems, demonstrating their effectiveness in solving such systems. Building upon this foundation, we utilize the divide-and-concur framework to address the real protein folding challenge. While protein folding involves complex thermodynamic processes that require a delicate balance of multiple physical forces, our constraint-based approach effectively approximates traditional force field models while offering significant computational advantages. The key innovation lies in incorporating interpretable constraints through the divide-and-concur framework, which systematically reduces the conformational search space. This enhancement enables ab initio folding of miniproteins on standard commercial computers within minutes, representing a substantial improvement in computational efficiency. The versatility of constraint-based methods extends beyond protein folding to other critical structural biology applications, particularly in multi-conformer refinement problems where configuration tangling phenomena arise. The modular nature of the divide-and-concur framework facilitates seamless integration of density constraints, making it particularly well-suited for multi-constraint optimization problems. This work establishes constraint-based approaches as powerful tools for tackling some of the most computationally demanding problems in modern structural biology.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Physics
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Physics
Grantor
Cornell University
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Mandaiya, Avinash
Chair dc:contributor.chair
  • Elser, Veit
Committee members dc:contributor.committeemember
  • Maxson, Jared
  • Myers, Christopher

Rights

dc:rights
Statement dc:rights
  • Attribution-ShareAlike 4.0 International
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
17423366
Dc Identifier Other
ProQuest Submission ID: 15288
ProQuest Publication ID: 32243434

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Mandaiya, Avinash. Constraint-Based Methods for Macromolecular Structure Prediction and Multi-Conformer Refinement. Doctor of Philosophy thesis, Cornell University, 2025. https://hdl.handle.net/1813/121111