Development of data processing methods for high resolution mass spectrometry-based metabolomics with an application to human liver transplantation

Abstract

Direct Infusion (DI) Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry (MS) is becoming a popular measurement platform in metabolomics. This thesis aims to advance the data processing and analysis pipeline of the DI FT-ICR based metabolomics, and broaden its applicability to a clinical research. To meet the first objective, the issue of missing data that occur in a final data matrix containing metabolite relative abundances measured for each sample analysed, is addressed. The nature of these data and their effect on the subsequent data analyses are investigated. Eight common and/or easily accessible missing data estimation algorithms are examined and a three stage approach is proposed to aid the identification of the optimal one. Finally, a novel survival analysis approach is introduced and assessed as an alternative way of missing data treatment prior univariate analysis. To address the second objective, DI FT-ICR MS based metabolomics is assessed in terms of its applicability to research investigating metabolomic changes occurring in liver grafts throughout the human orthotopic liver transplantation (OLT). The feasibility of this approach to a clinical setting is validated and its potential to provide a wealth of novel metabolic information associated with OLT is demonstrated.