Exploring heterogeneity in respiratory disease through systems biology A precision medicine approach

In this thesis the heterogeneity of respiratory diseases such as moderate-to-severe asthma and long COVID was explored using machine learning techniques. Different layers of information were used, including clinical characteristics, the blood and nasal transcriptome, the microbiome and the epigenome to uncover broad aspects of the diseases. The objectives of the studies included in thesis were to 1) cluster patients with similar disease pathophysiologies, 2) identify pathways or disease mechanisms underlying long COVID and 3) identify biomarkers for potential treatment or identification for a personalized medicine approach.
During this work we able to gather more insight into the disease pathology of both diseases and found two targets that might be suitable for a personalized intervention strategy in long COVID, namely butyrate producing bacteria in the gut microbiome and SMURF1 expression in the nasal epithelium. Additionally, we showed that the heterogeneity in asthma required the use of more advanced machine learning techniques to discover meaningful differences in the microbiome between uncontrolled and controlled asthmatics, and connected different molecular layers to identify a potential regulatory role for two taxa, highlighting the benefit of combining multiple layers of information in heterogeneous diseases. The works in this thesis provide a start in the characterization of long COVID and the use of advanced machine learning methods in respiratory research.