Automatic peak detection coupled with multivariate curve resolution–alternating least squares for peak resolution in gas chromatography–mass spectrometry

Abstract

Gas chromatography–mass spectrometry (GCMS) has been extensively used in complex sample analysis for the high-throughput characterization of volatile and semivolatile compounds. However, the accurate extraction of compound information remains challenging. Here, we present a combined algorithm strategy for GCMS data analysis to accurately screen metabolites across groups. First, chromatographic peaks in a total ion chromatogram (TIC) are extracted by using a Gaussian smoothing strategy and aligned on the basis of their mass spectra by a dynamic programing algorithm. The aligned TIC peaks are then registered into a component list table by applying a nearest-neighbor clustering algorithm. Significantly expressed TIC peaks among groups are screened through statistical analysis, such as ANOVA. Second, a chemometric method of multivariate curve resolution–alternating least squares for the peak resolution of the screened TIC peaks is utilized to retrieve the chromatographic and mass spectral profiles of coeluted components. The developed strategy is employed for the analysis of standard and complex plant sample datasets. Results indicate that our methodology is comparable with several state-of-the-art methods that are widely used in GC–MS-based metabolomics.