高级检索
    陈红丽, 周航, 杨永锋, 胡静宜, 刘茂林, 刘向真, 贾涛. 基于游离氨基酸的烤烟配打模块香型判别方法研究[J]. 中国烟草科学, 2023, 44(5): 79-85. DOI: 10.13496/j.issn.1007-5119.2023.05.010
    引用本文: 陈红丽, 周航, 杨永锋, 胡静宜, 刘茂林, 刘向真, 贾涛. 基于游离氨基酸的烤烟配打模块香型判别方法研究[J]. 中国烟草科学, 2023, 44(5): 79-85. DOI: 10.13496/j.issn.1007-5119.2023.05.010
    CHEN Hongli, ZHOU Hang, YANG Yongfeng, HU Jingyi, LIU Maolin, LIU Xiangzhen, JIA Tao. Identification Methods for Flue-cured Tobacco Flavor in Blending and Threshing Module Based on Free Amino Acids[J]. CHINESE TOBACCO SCIENCE, 2023, 44(5): 79-85. DOI: 10.13496/j.issn.1007-5119.2023.05.010
    Citation: CHEN Hongli, ZHOU Hang, YANG Yongfeng, HU Jingyi, LIU Maolin, LIU Xiangzhen, JIA Tao. Identification Methods for Flue-cured Tobacco Flavor in Blending and Threshing Module Based on Free Amino Acids[J]. CHINESE TOBACCO SCIENCE, 2023, 44(5): 79-85. DOI: 10.13496/j.issn.1007-5119.2023.05.010

    基于游离氨基酸的烤烟配打模块香型判别方法研究

    Identification Methods for Flue-cured Tobacco Flavor in Blending and Threshing Module Based on Free Amino Acids

    • 摘要: 为明确不同香型烤烟配打模块游离氨基酸组分差异,并实现对清香型、中间香型及浓香型烤烟复烤模块的准确分类,选取7个清香型、7个中间香型和8个浓香型烤烟配打模块为研究对象,对其游离氨基酸含量进行OPLS-DA分析,结合SVM模型进行香型判别。结果表明:不同香型烤烟模块游离氨基酸总量范围为13 141.72~27 695.47 μg/g,其中脯氨酸和天冬酰胺含量在各香型中占比最高,天冬氨酸、谷氨酸、甘氨酸、丙氨酸、胱氨酸、亮氨酸、酪氨酸、苯丙氨酸、4-氨基丁酸、赖氨酸、组氨酸、色氨酸、精氨酸和游离氨基酸总量在不同香型间存在显著性差异。丙氨酸和天冬氨酸在OPLS-DA分析中VIP值大于1且在不同香型间差异显著(p<0.05),可筛选为香型判定指标。基于21种游离氨基酸的前两种主成分构建的SVM模型对模块香型的判别率准确率为86.36%;基于筛选出的2个游离氨基酸特征指标构建的SVM模型判别准确率为90.91%。基于丙氨酸和天冬氨酸的SVM模型判别率和可信度都在较高水平,可应用于烤烟配打模块的香型判别。

       

      Abstract: Aiming to clarify the differences of free amino acid components in different flavor types of flue-cured tobacco blending and threshing modules, and classify redrying modules of fresh flavor, neutral flavor and robust flavor types, the blending modules of 7 fresh, 7 neutral and 8 robust flavor types were selected, and their free amino acid contents were determined by OPLS-DA analysis, flavor types were identified by SVM model. Results showed that total free amino acids in different flavor types of flue-cured tobacco ranged from 13 141.72 to 27 695.47 μg/g, and the contents of proline and asparagine exhibited the highest among all flavor types. Significant differences among different flavor types occurred in the total amount of aspartate, glutamic acid, glycine, alanine, cystine, leucine, tyrosine, phenylalanine, 4-aminobutyric acid, lysine, histidine, tryptophan, arginine and free amino acid. In OPLS-DA analysis the VIP values of alanine and aspartic acid were above 1 and significantly different among various flavor types (p<0.05), hence could be selected as evaluation indexes. The SVM model based on the first two principal components of 21 free amino acids exhibited an accuracy of 86.36%. The discriminant accuracy of SVM model based on two free amino acid characteristic indexes reached 90.91%. The SVM model based on alanine and aspartic acid exhibited high discrimination rate and reliability, and can be applied in the flavor discrimination of flue-cured tobacco blending modules.

       

    /

    返回文章
    返回