DOI
https://doi.org/10.25772/Y053-MY89
Author ORCID Identifier
https://orcid.org/0000-0002-2743-3637
Defense Date
2023
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Pharmaceutical Sciences
First Advisor
Dayanjan Shanaka Wijesinghe
Abstract
Heart failure with reduced ejection fraction (HFrEF) is a highly prevalent clinical syndrome with a mortality rate worse than most cancers, despite the availability of numerous pharmacotherapies and medical devices for management. Imperfect clinical biomarkers have been suggested as a reason for suboptimal therapy, and biomarker research has revealed an overactive innate immune response and inflammation, driven by overexpression of interleukin-1 (IL-1), as an undermedicated pathophysiological pathway. The metabolome offers a source for pharmacodynamic biomarkers that can capture the residual risks missed by conventional markers. However, current methodologies for identifying metabolomic biomarkers are limited, as the annotation of compounds from untargeted mass spectrometry assays typically relies on incomplete spectral libraries. Therefore, quantitative structure retention relationship (QSRR) models, which derive a relationship between retention time (RT) with the physicochemical properties of known compounds, can help annotate metabolomics data. In this dissertation, a natural language processing tool, LRN, is introduced to expedite systematic literature reviews. Central to this dissertation are two studies: in the first study, 20 pharmacodynamic biomarkers are developed for an IL-1 receptor antagonist, anakinra, with ribonic acid determined as a metabolomic therapeutic target for anakinra. Additionally, 3 anakinra-specific phenotypic metabolic pathways elucidated mechanisms underlying HFrEF. The second study employed an artificial intelligence workflow, MetaboLyte, that deployed 6 QSRR machine learning algorithms to identify 27 previously unknown lipids from metabolomics studies. MetaboLyte also distinctly enhanced QSRR modeling for carboxylic acids by predicting RTs at deeper taxonomic levels. The results presented herein aid in the advancement of precision medicine for HFrEF.
Rights
© Joshua M. Morriss
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
4-26-2023
Included in
Artificial Intelligence and Robotics Commons, Bioinformatics Commons, Cardiovascular Diseases Commons, Cardiovascular System Commons, Computational Biology Commons, Computational Chemistry Commons, Data Science Commons, Immune System Diseases Commons, Other Pharmacy and Pharmaceutical Sciences Commons