Please use this identifier to cite or link to this item:
|Title:||Computational modelling and data analysis on the virulence of influenza viruses||Authors:||Zhou, Xinrui||Keywords:||Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
|Issue Date:||2019||Publisher:||Nanyang Technological University||Source:||Zhou, X. (2019). Computational modelling and data analysis on the virulence of influenza viruses. Doctoral thesis, Nanyang Technological University, Singapore.||Abstract:||Influenza virus, a rapidly evolving contagious virus causing seasonal flu, has been circulating globally for centuries. The first recorded influenza pandemic was in 1918, claiming at least 50 million lives. Until 1933, influenza viruses were isolated from human for the first time, proving that influenza is caused by a virus rather than a bacterium. Seasonal influenza has caused substantial social and economic burden worldwide, affecting school attendance, work absenteeism, industrial productivity, etc. It mainly infects the respiratory system and can cause pneumonia, severe complications, or deaths, especially among people at high risks. Flu vaccines have been designed as primary prevention to help defense viral infection in advance. However, the rapid and continuous evolution of influenza raises the challenge to prepare vaccine candidates matching the antigenicity of dominant circulating influenza viruses for the next season. Therefore, it is in urgent demand to characterize and predict the antigenicity of influenza in advance. Given the feasibility of high throughput sequencing techniques and enriched protein structure database, lots of computational models have been proposed to characterize antigenic properties of influenza. There have been many studies working on evolutionary models for tracing back the genomic variations and predicting the antigenic variants of influenza. However, current models for predicting the antigenicity are only applicable to one pre-defined subtype of influenza virus. The universal models for multiple-subtypes are still lacking. When it comes to virulence, the ability of the virus to cause disease among humans, it is a more complex problem involving the interaction with the immune system. From the medical perspective, the virulence level of influenza viruses is measured with the severity of an infection, the capability of drug resistance and transmission among hosts. There are still no consistent measurements for quantifying the virulence level of an influenza viral strain. The objective of this dissertation is to construct computational models for profiling the virulence of influenza viruses. In this dissertation, the virulence level is quantified from the virus perspective only, including the sequence analyses on the genomic variation, and structural analyses on the receptor binding. The proposed sequence models for genomic variation include a phylogenetic-tree based method for pairwise co-mutations of influenza intra-proteins, and a sequential rule mining based approach for co-occurring mutations at multiple sites, even on different proteins. For profiling the receptor binding specificity, a structure-based model was proposed to characterize the binding modes between the influenza viral membrane protein (HA) and the human receptors. Both sequence models and structural models are integrated into a pipeline to quickly profile the virulence of influenza viral strains. Results of this proposed pipeline on our newly sampled influenza viral strains among outpatients and inpatients in Singapore highlighted viral subtypes and strains that are more infectious or pathogenic, which are consistent with the local observations.||URI:||https://hdl.handle.net/10356/137781||DOI:||10.32657/10356/137781||Rights:||This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).||Fulltext Permission:||open||Fulltext Availability:||With Fulltext|
|Appears in Collections:||SCSE Theses|
Updated on Jan 29, 2023
Updated on Jan 29, 2023
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.