摘要
基于分子形状指数(mK)表征20种食品香味物质的分子结构.利用最佳变量子集回归方法分别建立食品香味物质的气相色谱保留指数(IR)、香味阈值(XT)与mK的二元数学方程,其传统相关系数分别为0.998和0.976、并逐一剔除法的交互检验系数(q2)分别为0.991和0.684.所建数学模型不仅揭示了影响IR和XT的主要因素,而且具有良好的预测能力与稳健性.
Kier＇s molecular shape indices（mK） were used to describe the molecular structures of 20 flavoring substances in this paper. The quantitative structure-activity relationships（QSAR） between gas chmmatography retention indices（/R）, threshold value of flavor（XT） and two-variable （^1K,^2K） of flavoring substances was set up using a Leaps-and-Bounds Regression （LBR）. The two models of two parameters （such as ^1K, 2K）, which the calibrated correlation coefficients（ R ） was 0.998,0. 976, respectively, and the validated correlation coefficients（Q2） of the Leave- One-Out（LOO） was 0. 991,0. 684, respectively. The two-variable models has shed light on some main structural fac- tors influencing the IR, XT of flavoring substances, which showed the good estimation ability and robustness.
出处
《分子科学学报》
CAS
CSCD
北大核心
2011年第5期367-370,共4页
Journal of Molecular Science
关键词
食品香味物质
分子形状指数
气相色谱保留指数
香味阈值
定量构效关系
flavoring substance
molecular shape indices
gas chromatography retention indices
threshold value of flavor
quantitative structure-property/activity relationship
