The Identification Of Cold-Hot Medicine Property (CHMP)and The Establishment Of Partial Least Squares Path Model(PLS Path Model) Based On Primary Materials

Objective:To identify the Cold-Hot medicine property (CHMP) by measuring the contents and HPLC, GC-MS chromatograms of primary materials of60kinds of herbs and using PAST software for establishing the discriminant function; to identify the Cold-Hot medicine property markers (CHMP-markers) by screening the best statistical identification model based on the primary materials; to establish partial least squares path model (PLS path model) between the primary materials and CHMP by SmartPLS2.0software for investigating the primary materials on impact of medicine property and the ralationship between the primary materials.Methods:The total protein contents were determined by Coomassie brilliant blue method, the amino acids contents were determined by UV spectrophotometric method, the tryptophan contents were determined by pre-column derivatization method, and the data were collected for analyzing by Fisher method in PAST software, then the discrimination function was established based on the protein and amino acids to identify the CHMP; the sugar contents were determined by Felling method, the total polysaccharide contents were determined by sulfuric acid-anthrone colorimetric method, the monosaccharide contents were determined by alkali copper sulfate methods, and the data were analyzed by Fisher method and established the discrimination function based on the carbohydrate contents to identify the CHMP; the GC-MS chromatograms of polysaccharide and monosaccharide were determined, and the data were also analyzed by Fisher method and established the discrimination function based on GC-MS chromatograms of polysaccharide and monosaccharide to identify the CHMP; the GC-MS chromatograms of lipids were determined, and the data were analyzed by Fisher method and established the discrimination function based on GC-MS chromatograms of lipid to identify the CHMP; the data of primary materials contents and primary materials chromatograms were collected respectively, and the Fisher function was established based on the contents and chromatograms to identify the CHMP; screened the best statistical identification model to identify the CHMP-markers; used SmartPLS2.0software to establish PLS path model.Results:The identification correct rate was81.70%of the Fisher function based on protein and amino acid contents; the identification correct rate was88.33%of Fisher function based on carbohydrate contents; the correct rates were88.33%and93.33%of Fisher function based on GC-MS of polysaccharides and monosaccharide; the correct rate was96.67%of Fisher function based on GC-MS of lipids; the identification correct rate was86.70%of discriminant function based on primary materials contents and88.33%of discriminant function based on primary materials chromatograms. The best statistical identification model for GC-MS of polysaccharide was PLS-DA, the best model for GC-MS of monosaccharide was PLS-DA or PCA-DA, the best model for GC-MS of lipids and HPLC of amino acid was SVM. The PLS path the model results showed that:amino acids, type2of carbohydrates, type3of carbohydrates and type2of lipids had positively related to the medicine property while type1of carbohydrates, type4of carbohydrates, type1of lipids, type3of lipids, type4of lipids and type5of lipids had negative correlation with the medicine property, the R square value of medicine property was0.398, which suggested that10latent variables had39.8%of explanatory power for the medicine property.Conclusion:The discriminant function established by Fisher method based on primary materials has good identification ability, which suggests that the CHMP can be identified by statistical methods, CHMP-markers of primary materials could be preliminary identified by the best screened statistical identification model, and it initially confirms that the CHMP has certain correlation with primary materials.