Comparison of Pre-Processing and Variable Selection Strategies in Group Based GCxGC-TOFMS Analysis
Paulina K. Piotrowski, Benjamin J. Place
Chemometric analysis of comprehensive two-dimensional chromatography coupled to time-of-flight mass spectrometry data has been reported with various workflows. The report presented herein aims to investigate the effects of different pre-processing and variable selection strategies on the scores plot outputs from GCxGC-TOFMS data acquired from lavender and tea tree essential oils. Our results suggest that pre-processing such as applying log transformation to the data set can result in significant differentiation of sample clustering when compared to only mean centering. Additionally, exploring differences between unsupervised (ANOVA and F-ratio) and supervised (PLS-DA) feature selection resulted in little variation in scores plots. This work highlights the effects different chemometric workflows can have on results to help facilitate harmonization efforts.
and Place, B.
Comparison of Pre-Processing and Variable Selection Strategies in Group Based GCxGC-TOFMS Analysis, Separations, [online], https://doi.org/10.3390
(Accessed February 21, 2024)