ORCID ID
Graduation Date
Spring 5-4-2024
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Programs
Biomedical Informatics
First Advisor
Dr. Chittibabu (Babu) Guda
Abstract
The global outbreak of COVID-19, triggered by the novel coronavirus, Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), has spurred an urgent need for a deeper comprehension of the molecular mechanisms involved in the host-virus interactions. Despite advancements in transcriptomic technology and computational resources, limited attention has been given to the holistic integration of molecular and clinical data to characterize the genotype/phenotype aspects of the disease.
This study analyzes gene expression patterns in various tissues, including the lung, nasal, blood, and placenta, in patients with COVID-19 to identify differentially regulated genes and pathways. We also evaluated organ-specific gene co-expression patterns that revealed the functional relationships and interactions among genes, along with potential tissue-specific biomarkers such as APLNR and BPIFB1 in Lung and A2MP1 and AATK in the blood. This analysis helped to understand the tissue-level responses and provide insights into why specific organs are more susceptible to infection than others. Further, we evaluated different Machine Learning (ML) models along with the integration of gene expression data, clinical features, and co-morbidity data for predicting COVID-19 severity. The XGBoost, with 95% accuracy, outperformed other methods, including Logistic Regression, XGBoost, Naïve Bayes, and Support Vector Machine. SHAP analysis provided the most discriminative features, including COX14, absolute neutrophil count, and viremia, which paved the way to understanding the patient’s severity level.
These findings highlight integrating clinical, co-morbidity, and gene expression data to predict the severity of COVID-19 and offer valuable prognostic insights for clinicians to optimize treatment strategies.
Recommended Citation
Sethi, Sahil, "Gene Co-Expression and Machine Learning Approaches to Compare SARS-CoV-2 Infected Tissues in Humans" (2024). Theses & Dissertations. 823.
https://digitalcommons.unmc.edu/etd/823
Comments
2024 Copyright, the authors