Automation of parallel factor analysis (PARAFAC) for peak resolution in GC x GC-TOFMS data.

摘要

Instrumentation performing two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC x GC-TOFMS) is a powerful analytical technique capable of increased resolution in comparison with traditional gas chromatography (GC) or gas chromatography coupled with mass spectrometry (GC-MS). Data produced by GC x GC-TOFMS can be arranged into meaningful three or higher-way arrays that allow application of advanced chemometric techniques offering higher-order advantages. One such technique is parallel factor analysis (PARAFAC), which can be used to mathematically separate overlapping peaks or mixed signals in GC x GC-TOFMS data. However, application of PARAFAC is complicated by the need to use PARAFAC models having appropriate numbers of factors for the signals to be resolved. Models are typically evaluated by an analyst to determine if they are appropriate for the data i.e. an appropriate number of factors has been used to resolve the signals from analyte(s) of interest or whether new models with more or less factors need to be created. Automated methods could reduce analyst burden and increase objectivity in this process. To address this problem, several newly developed automated methods for performing PARAFAC and determining which PARAFAC models are appropriate in different analysis situations for GC x GC-TOFMS data are presented and evaluated in this dissertation, and future prospects and directions for automation in the context of GC x GC-TOFMS data analysis are discussed.