A modification on the vector cosine algorithm of Similarity Analysis for improved discriminative capacity and its application to the quality control of Magnoliae Flos

•A modified algorithm was proposed based on the original Similarity Analysis method.•The improved discriminative capacity was illustrated in theory and practice.•The advantages were confirmed by the results of quantitative analysis.

Chromatographic fingerprint analysis has been widely used in quality control of herbal medicines, and Similarity Analysis (SA) as a well-established method has been applied in the quality control practice as well as publications related to the study of herbal medicines and preparations. However, in some cases the results of SA do not fit well with those of other chemometric approaches and quantitative analysis, and the problem remains unsolved. In this study, a modified SA algorithm has been proposed, with its advantages discussed in theory. The extract of dried flower bud of Magnolia biondii Pamp. obtained by pressurized solvent extraction was then selected as a case to verify the modified algorithm. After identification of the components, fingerprint analysis was performed using different chemometric methods including Hierarchical Cluster Analysis (HCA), Principal Component Analysis (PCA) as well as original and modified SA methods, and the improved discriminative capacity of modified SA algorithm was illustrated. Characteristic chemical markers were then identified using the modified SA approach and then confirmed using PCA method. The quantitative results were then utilized to confirm the advantage of modified SA approach over the original one. The study made a modification to the widely applied SA algorithm, which was possibly a beneficial improvement in fingerprint analysis and quality control practice of herbal medicines.