Advanced Theoretical Modeling Approaches and Innovative Data Processing Workflows in GC×GC-ToF-MS

Comprehensive two-dimensional gas chromatography coupled with mass spectrometry
(GC×GC-MS) is nowadays widely recognized as an indispensable analytical method
across multiple research fields. The research presented in this thesis aims to comprehensively understand GC×GC-MS by addressing both method development and data processing. The first part of this research work is devoted to developing a novel approach to retention time prediction called the “Top-down Approach”. Departing from previous modeling works, this approach offers a new perspective on retention time modeling, enhancing method optimization and separation space description and visualization. This approach considers the GC×GC system as two physically separate one-dimensional separations and combines separate retention time predictions for each dimension to describe the two-dimensional chromatographic separation space. The model demonstrated good agreement with experimental data and exhibited potential as a system-independent method. The second part of this work introduces innovative data processing strategies to maximize information extraction. Advanced machine learning algorithms, with a particular focus on random forest (RF), are explored due to their effectiveness in handling large amounts of data. A new data processing technique termed the “Tile-based RF Analysis” is introduced. It combines the tile-based approach with RF for analyte discovery and feature importance assessment. Additionally, a hybrid data processing workflow that combines traditional statistical methods with RF is developed and applied to food flavor analysis by combining both targeted and untargeted analyses to differentiate between regular
and decaffeinated coffee samples.