Using host “omic” datasets to better understand and diagnose paediatric infectious and inflammatory diseases
File(s)
Author(s)
Jackson, Heather
Type
Thesis or dissertation
Abstract
The presence of an infectious or inflammatory disease leads to changes in gene expression and protein abundance which are detectable in the host’s blood. Through studying the host omic profiles of individuals with infectious and inflammatory diseases, we can identify molecular changes that occur due to disease. Sparse combinations of genes or proteins that distinguish between disease groups of interest can be determined using feature selection. These sparse combinations, which are known as host diagnostic signatures, can be developed into point-of-care tests for rapid and accurate disease diagnosis. Host omic profiles can also be used to uncover important genes, proteins, and biological pathways implicated in disease progression and manifestation.
In this thesis, I have used host transcriptomic and proteomic profiles to address three major challenges in paediatric infectious and inflammatory disease. I have firstly described a 6-protein host diagnostic signature that can accurately differentiate between bacterial and viral infections in febrile children, with area under the receiver operating characteristic curves (AUCs) between 89.4%-93.6% in an independent validation set. Secondly, Kawasaki disease (KD) is a paediatric inflammatory syndrome without a known cause. I have used host transcriptomic and proteomic profiles to determine whether the host response to KD detectable on the molecular level has more in common with the host response to bacterial or viral infections, or indeed neither. These analyses have revealed considerable heterogeneity in the host response to KD. Finally, I have discovered a 5-gene host diagnostic signature that accurately distinguished multisystem inflammatory syndrome in children from KD, bacterial infections, and viral infections with AUCs between 91.7%-93.2% in an independent validation set.
In summary, this thesis has demonstrated that the combination of high-dimensional omic datasets and statistical and machine learning tools provides invaluable insights into the biology of paediatric infectious and inflammatory diseases and offers alternative diagnostic approaches.
In this thesis, I have used host transcriptomic and proteomic profiles to address three major challenges in paediatric infectious and inflammatory disease. I have firstly described a 6-protein host diagnostic signature that can accurately differentiate between bacterial and viral infections in febrile children, with area under the receiver operating characteristic curves (AUCs) between 89.4%-93.6% in an independent validation set. Secondly, Kawasaki disease (KD) is a paediatric inflammatory syndrome without a known cause. I have used host transcriptomic and proteomic profiles to determine whether the host response to KD detectable on the molecular level has more in common with the host response to bacterial or viral infections, or indeed neither. These analyses have revealed considerable heterogeneity in the host response to KD. Finally, I have discovered a 5-gene host diagnostic signature that accurately distinguished multisystem inflammatory syndrome in children from KD, bacterial infections, and viral infections with AUCs between 91.7%-93.2% in an independent validation set.
In summary, this thesis has demonstrated that the combination of high-dimensional omic datasets and statistical and machine learning tools provides invaluable insights into the biology of paediatric infectious and inflammatory diseases and offers alternative diagnostic approaches.
Version
Open Access
Date Issued
2022-07
Online Publication Date
2024-06-30T23:01:30Z
2024-08-09T13:52:02Z
Date Awarded
2023-01
Copyright Statement
Creative Commons Attribution NonCommercial Licence
Advisor
Kaforou, Myrsini
Herberg, Jethro
Levin, Michael
Sponsor
Wellcome Trust (London, England)
Grant Number
215214/Z/19/Z
Publisher Department
Department of Medicine
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)