Metabolomics and biosensor approaches to the detection of fever associated diseases

Nastase, Ana-Maria (2022) Metabolomics and biosensor approaches to the detection of fever associated diseases. PhD thesis, University of Glasgow.

Abstract

Febrile illnesses are still a major cause of mortality and morbidity globally and the failure to detect and correctly diagnose a specific disease associated with fever is partly responsible for this. This thesis aimed to investigate a biosensor-based method for the detection of fever associated diseases and to further explore the molecular mechanisms and possible biomarkers of febrile illnesses by employing a metabolomics-based approach. The biosensor platform is based on a complementary metal oxide semiconductor technology, which has both technological and economic advantages. Due to the small size of the microchip, accurate signal processing becomes challenging and, thus, computational methods were developed and tested for the quantitative detection of antibodies in a solution tested on the biosensor platform. Three methods, one based on a deterministic approach and two others based on machine learning (ML) algorithms, were tested and compared for the detection of a reaction spot intensity using synthetically generated images. Next, in order to develop an immunoassay protocol for the detection of one specific fever associated infectious disease, human African trypanosomiasis (HAT), several steps were taken. First of all, a suitable and sensitive method of detection was selected, i.e. enzyme linked immunosorbent assay (ELISA). Next, four recombinant antigens currently used for the detection of HAT were selected based on previous evidence and developed using molecular cloning techniques in E.coli. These were tested on infected and control humans serum samples obtained from endemic regions of the Democratic Republic of Congo (DRC). Disposable poly-methyl methacrylate (PMMA) slides which were chemically functionalised were used on top of the chip as the immunoassay surface. Titrations for the selected antigens/antibody were tested using an indirect ELISA-like protocol and the best results after fitting a calibration curve were obtained for an antigen concentration of 2.5 µg/ml. The detection of the antibody to the trypanosome antigen invariant surface glycoprotein 65 (ISG65) proved to be successful and the protocol could be replicated for all the other antigens. However, technical challenges and the closure of the laboratory during the Covid-19 pandemic precluded my taking this part of the project to its conclusion. Following this, metabolomics datasets studying disparate febrile infectious illnesses obtained using liquid chromatography coupled to mass spectrometry (LC-MS) were used in order to investigate and detect possible metabolite-based biomarkers common to fever-associated diseases. A warping based method was developed in order to enable integration by alignment of disparate LC-MS metabolomics datasets. Integration was performed by correcting the RT drift between the datasets using fitted Gaussian Process regression models, a supervised ML method, which was followed by direct matching alignment using MZmine2. The correction was performed by using the standard reference mixture (SRM) information. Statistical analysis on the meta-dataset was performed using linear modelling implemented in the limma R-package. Comparison was made between infected and control samples and commonality was established using the fold change values obtained for the individual datasets. Annotation was carried out by matching the compounds against metabomlomics datasets and through mummichog software, which was also used for pathway analysis. The features obtained from this analysis which were putatively annotated were classified into categories (amino acids, sugars, lipids, nucleotides, etc.). Features in common to all datasets were used to make a connection to the previously established molecular basis of fever. Significant changes were identified to several metabolic pathways, with the most notable perturbations being within the kynurenine pathway, a branch of tryptophan metabolism. Also, features specific to each dataset were used to evaluate the accuracy of the fever biomarkers and investigate possible biomarkers for each different fever-associated disease.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: Computational biology, machine learning, metabolomics, bayesian inference, lab-on-a-chip, biochemistry, biomedical engineering.
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science; Q Science > QR Microbiology > QR180 Immunology; R Medicine > R Medicine (General)
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Infection & Immunity
• Funder's Name: Engineering and Physical Sciences Research Council (EPSRC)
• Supervisor's Name: Barrett, Professor Michael, Rogers, Dr. Simon and Cumming, Professor David
• Date of Award: 2022
• Depositing User: Theses Team
• Unique ID: glathesis:2022-82843
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 04 May 2022 08:49
• Last Modified: 16 May 2022 13:11
• Thesis DOI: 10.5525/gla.thesis.82843
• URI: https://theses.gla.ac.uk/id/eprint/82843