Compositional profiling and classification of radix Angelicae sinensis by effective GC-IMS and chemometric tool

A new analytical architecture was proposed in this study to profile the volatile fraction of edible plant materials of radix Angelicae sinensis (RAS) using non-targeted gas chromatography-ion mobility spectrometry (GC-IMS) technique. High dimensional GC-IMS data were effectively tackled in two encompassing stages and three exploratory spaces by a pool of chemometric methods. 33 flavor compounds were remarkably identified in 287 RAS batches sourced from different geographic origins. The data augmentation through competitive adaptive reweighted sampling method was beneficial to discerning significant features in feature space. The samples were classified in sample space up to a high accuracy of 99.37% by principal discriminant variate regime. The use of augmented features to map sample patterns exhibited a significant edge over peak features automatically detected in LAV software. The work provided a judicious strategy to address the most commonly encountered problems in the GC-IMS analytical context of volatile organic compounds of natural materials. • GC-IMS was used to profile the VOCs of RAS materials and 33 compounds identified. • Architecture was proposed for GC-IMS data analysis in two stages and three spaces. • Data augmentation benefited to improve the performance of classification model. • Three-class samples were discriminated by PDV model as per producing origins.