Construction of Comprehensive Flavonoid Analysis Tool by Using UV‐vis Spectra Library, In‐house Database, and Chemometrics

Liquid Chromatography and mass spectrometry methods, especially ultra-high performance liquid chromatography coupled with high resolution accurate mass-mass spectrometry (UHPLC-HRAM-MS), have become the best methods for flavonoid identification and quantification. However, processing acquired UHPLC-HRAM-MS data for flavonoid analysis is very challenging and highly expertise-dependent because of the complexity of the physical and chemical properties of the flavonoids. An expert data analysis program, FlavonQ, has been developed to facilitate this process. The program first categorizes the flavonoids using a chemometric model based on the UV-Vis spectra library compiled for 146 flavonoid reference standards. A novel stepwise classification strategy is used that provides data representation in each step as optimized by a projected distance resolution (PDR) method. The stepwise classification strategy significantly improves the performance of the classifiers which results in more accurate and reliable classification results. An in-house flavonoid database which contains 5686 previously reported flavonoids is used for identification of flavonoids. FlavonQ was validated by analyzing data from samples with spiked flavonoid mixed standards and plant samples including blueberry, mizuna, purple mustard, red cabbage, and red mustard green extract. Accuracies for identification for all samples were above 88%. FlavonQ greatly facilitates the identification and quantitation of flavonoids from UHPLC-HRAM-MS data. The process is automated, saving tremendous resources, and allowing less-experienced people to perform data analysis on flavonoids with reasonable results. Support or Funding InformationThis research is supported by the Agricultural Research Service of the U.S. Department of Agriculture, an Interagency Agreement with the Office of Dietary Supplements at the National Institutes of Health (Y01 OD001298-01). The John A. Milner Fellowship program by USDA Beltsville Human Nutrition Research Center and the NIH Office of Dietary Supplements is acknowledged for the support to Dr. Mengliang Zhang. We thank to Dr. Peter de B. Harrington for providing his codes on chemometric models.