Constraint Satisfaction for NMR Analysis

Abstract

This thesis presents a framework for molecular structure elucidation through integrating Nuclear Magnetic Resonance (NMR) data and constraint satisfaction techniques. The research addresses the challenge of generating molecular structures by employing a constraint satisfaction framework incorporating constraints derived from NMR spectroscopy, chemical principles and forbidden structures representing solutions as molecular structures. These constraints are used to represent solutions as molecular multigraphs. We begin by providing an overview of the research landscape, emphasising the significance of molecular structure elucidation across various fields. Our implementation enables the identification of plausible molecular structures by exploring the space of variable assignments while satisfying predefined constraints. Additionally, we introduce the theoretical foundations of the constraint satisfaction framework employed in this research. We conduct a detailed examination of constraint-based methods and their applicability to molecular structure generation. Developing a generic constraints framework is the foundation of our methodology, facilitating the integration of diverse sources of information and enhancing adaptability. Moreover, we present the formalism of the constraints utilised in the framework for accurate structure generation. We proceed to demonstrate the implementation of the constraint satisfaction framework and its application to challenging NMR problems. Through comprehensive case studies and experimentation, we demonstrate the effectiveness of our approach in generating precise and efficient solutions, highlighting its utility in real-world scenarios. Additionally, we utilise the constraint satisfaction framework to establish NMR’s inherent ambiguity in molecular structure elucidation. Especially when there are multiple possible interpretations or when certain constraints are insufficient to uniquely determine the structure.