DEVELOPMENT OF COMPUTATIONAL METHODS FOR MASS SPECTROMETRY-BASED UNTARGETED METABOLOMICS DATA ANALYSIS

Abstract

Mass spectrometry coupled with liquid or gas chromatographic separation has been widely used in metabolomics. However, there are still limitations in the current data processing pipelines. In this thesis, we first addressed the limitations in MS1 based analysis and developed a software package “MetTailor” containing two novel post-alignment data preprocessing functions: 1) re-align the potential misaligned peaks; 2) normalize data to adjust the temporal variation along RT. Next, we developed a data processing framework “MetaboDIA” to addressed the limitations in MS/MS-based analysis for both spectral library construction and DIA-MS extraction. In “MetaboDIA”, we first assigned putative molecular formulae to MS1 precursor and constructed consensus MS/MS spectral assay library based on the reproducibility with rigorous quality control steps(library can be built from DDA or DIA data). The library was then used to perform extraction of transition-level peak intensities from DIA data.