Dissertation
HIGH-THROUGHPUT CALCULATION OF SMALL MOLECULE NMR CHEMICAL SHIFTS AND MOLECULAR FRAGMENTS
Washington State University
Doctor of Philosophy (PhD), Washington State University
01/2021
DOI:
https://doi.org/10.7273/000006431
Handle:
https://hdl.handle.net/2376/119009
Abstract
The majority of metabolites are not available as authentic reference material, making confident identification of molecules in complex samples unachievable. Despite recent technological advances, the comprehensive, confident and high-throughput identification of compounds detected in metabolomics studies through currently available experimental methods represents a major roadblock for both technical and economic reasons. In silico NMR libraries are an attractive alternative to the reliance on reference libraries constructed by analysis of authentic reference materials with limited commercial availability. In silico calculation of nuclear magnetic resonance (NMR) chemical shifts is a promising approach to expand available libraries and databases. Firstly, to address the limitations in metabolomics, a high-performance cheminformatics workflow, in silico Chemical Library Engine (ISiCLE), is introduced, designed to predict NMR chemical shifts of organic molecules by employing density functional theory (DFT). ISiCLE is evaluated on a set of molecules integrated with an error reduction approach (conformational sampling).
Secondly, the accuracy required for the identification of metabolites is investigated using 11,716 molecules. Of these molecules, 90% can be successfully identified in a pure sample when errors of 1H and 13C chemical shifts reach at least 0.6 ppm and 7.1 ppm, respectively. However, at that same level of accuracy, in complex mixtures, the level of accuracy for identification increases significantly.
Thirdly, an automated solvation module is designed to evaluate explicit and implicit solvation at varying levels of theory. It is applied to a set of carboxylic acids of which the chemical shifts are extremely sensitive to water-solvent effects and concluded 2.78 ppm and 0.47 ppm error for 13C and 1H chemical shifts, respectively.
Furthermore, an automated and efficient tool, substructure processing, enumeration, and comparison tool resource (SPECTRe), is introduced to provide an entire set of substructures for a given molecular structure. SPECTRe is validated with a set of 10,375 molecules.
This dissertation represents computational methods to address the limitations of small molecule identification and offers an alternative to generating NMR libraries experimentally by analyzing authentic reference materials. High-throughput and comprehensive metabolite identifications will revolutionize the understanding of the role of metabolites in environmental and biological samples.
Metrics
Details
- Title
- HIGH-THROUGHPUT CALCULATION OF SMALL MOLECULE NMR CHEMICAL SHIFTS AND MOLECULAR FRAGMENTS
- Creators
- Yasemin Yesiltepe
- Contributors
- Ryan Renslow (Advisor)Brigitte Ahring (Committee Member)Haluk Beyenal (Committee Member)Niranjan Govind (Committee Member)
- Awarding Institution
- Washington State University
- Academic Unit
- Chemical Engineering and Bioengineering, School of
- Theses and Dissertations
- Doctor of Philosophy (PhD), Washington State University
- Publisher
- Washington State University
- Number of pages
- 590
- Identifiers
- 99900592361801842
- Language
- English
- Resource Type
- Dissertation