Discrimination of Aurantii Fructus Immaturus and Fructus Poniciri Trifoliatae Immaturus by Flow Injection UV Spectroscopy (FIUV) and 1H NMR using Partial Least-squares Discriminant Analysis (PLS-DA)

Authors: ZAHANG, MENGLIANG - Ohio University; ZHANG, YANG - University Of Maryland; HARRINGTON, PETER DE - Ohio University; Chen, Pei

Submitted to: Analytical Letters
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/16/2015
Publication Date: 6/8/2015
Citation: Zahang, M., Zhang, Y., Harrington, P.B., Chen, P. 2015. Discrimination of Aurantii Fructus Immaturus and Fructus Poniciri Trifoliatae Immaturus by Flow Injection UV Spectroscopy (FIUV) and 1H NMR using Partial Least-squares Discriminant Analysis (PLS-DA). Analytical Letters. 49:711-722.

Interpretive Summary: Two simple fingerprinting methods, flow-injection ultra-violet spectroscopy (FIUV) and 1H nuclear magnetic resonance (NMR), for discrimination of Aurantii FructusImmaturus and Fructus Poniciri Trifoliatae Immaturususing were described. Both methods were combined with statistical analysis for the differentiation. The prediction rates of 100% were achieved using both data sets. A new validation set of data were collected by the FIUV method two weeks later and tested the model constructed by initial TACUV and AAUV data sets with no parameter changes. The classification rates were 95% and 100% with TACUV and AAUV data sets, respectively. Both FIUV and NMR methods are simple, high throughput and low-cost, which are good alternatives for discrimination studies.

Technical Abstract: Two simple fingerprinting methods, flow-injection UV spectroscopy (FIUV) and 1H nuclear magnetic resonance (NMR), for discrimination of Aurantii FructusImmaturus and Fructus Poniciri TrifoliataeImmaturususing were described. Both methods were combined with partial least-squares discriminant analysis (PLS-DA). In FIUV fingerprinting method, two data sets including total absorbance chromatogram of UV data sets (TACUV) and averaged UV absorbance data sets (AAUV) were constructed and evaluated. The prediction rates of 100% were achieved using both data sets by PLS-DA with leave-one-out cross-validation. The prediction rate from the NMR data by PLS-DA with leave-one-out cross-validation was also 100%. A new validation set were collected by FIUV two weeks later and predicted by PLS-DA models constructed by initial TACUV and AAUV data sets with no parameter changes. The classification rates were 95% and 100% with TACUV and AAUV data sets, respectively. Both FIUV and NMR methods are simple, high throughput and low-cost, which are good alternatives for discrimination studies.