摘要
利用化学法测定了粳稻精米粉必需氨基酸的含量,并分别建立了相应的近红外分析预测模型。结果表明,不同光谱预处理方法对近红外分析模型的预测结果有较大影响,采用光谱预处理的校正效果比不采用预处理的好。用偏最小二乘法(PLS)获得的粳稻精米粉缬氨酸、异亮氨酸、亮氨酸、苯丙氨酸、蛋氨酸、苏氨酸、赖氨酸等7种必需氨基酸含量的预测模型和交叉验证结果显示表明,最优校正决定系数(R2)和交叉检验均方误差(RMSECV)分别为0.8868、0.0303;0.8623、0.0237;0.9008、0.0359;0.8993、0.0278;0.5999、0.0256;0.76040、.0238;0.8543、0.0173。因此在水稻品质育种中,近红外光谱分析技术可用于除蛋氨酸、苏氨酸外的其余5种必需氨基酸含量的测定。
Using the value of essential amino acid contained in milled japonica rice powder analyzed by chemical method, the near infrared spectroscopy prediction models were set up. Results showed that there were obvious effects of different prespectral treatment on the prediction values of the near infrared spectroscopy analysis models, and the calibration effect was better using pre-spectral treatment than directly treatment. The prediction models derived from the partial least squares (PLS) and cross-certification for Valine, Isoleucine, Leucine, Phenylalanine, Methionine, Threonine and Lysine indicated that the optimal calibration determination coefficient (R^2) and cross-examination mean square errors (RMSECV) were 0.8868, 0.0303; 0.8623, 0.0237; 0.9008, 0.0359; 0.8993, 0.0278; 0.5999, 0.0256; 0.7604, 0.0238; 0.8543, 0.0173, respectively. In terms of the prediction models, it was proposed that except for Methionine and Threonine, other 5 kinds of essential amino acid could be chosen as targets for grain quality in rice breeding practice.
出处
《核农学报》
CAS
CSCD
北大核心
2007年第5期478-482,共5页
Journal of Nuclear Agricultural Sciences
基金
浙江省科技攻关项目(2004C12020
2005D70053
2005C22016)
关键词
近红外漫反射光谱法
粳稻
必需氨基酸
测定
near infrared diffuse reflectance spectroscopy
japonica rice ( Oryza sative L. )
essential amino acid
determination