近红外透反射光谱测定单粒稻种的蛋白质含量

Determination of Protein Content in Single Rice Kernels and Grains by Near-Infrared Transflectance Spectroscopy

为探索NIR光谱技术在水稻种子蛋白质含量分析中的应用,本研究细致分析了单粒稻种在不同光谱采集方式下的近红外光谱(NIRS)特征,并利用离子束诱变育种得到的水稻9311突变体库的种子,建立准确性较好的单粒糙米和单粒稻种的蛋白质定量模型。结果表明,与漫反射光谱采集方式下的单粒糙米蛋白质模型相比,透反射和透射光谱采集方式下能得到相关性较好的糙米蛋白质模型,其中单粒糙米蛋白质最优定量模型的决定系数(R^2)为0.97,预测均方根误差(RMSEP)为0.27%。在单粒稻种中,由于种壳的反射作用,漫反射光谱采集方式下依然无法建立准确性高的蛋白质模型,透反射光谱采集方式下能够建立具有一定预测能力的蛋白质定量模型(RMSEP=0.81%),透射光谱采集方式下能够建立准确性高的蛋白质定量模型(R^2=0.96,RMSEP=0.24%)。本研究结果为无损快速分析单粒稻种提供了一种解决方法。

To explore application of NIR spectroscopy in analysis of single rice seed protein content,we compared the near-infrared(NIR)spectral characteristics of rice seeds under different NIR measurement modes,and then established quantification models with relative good accuracy for protein content in single rice kernels and rice grains by using the seeds from the ion-beam-induced indica rice cultivar 9311 mutant library.Our results revealed that,for the assessment of protein content in single brown-rice kernels,the quantitative models based on the transflectance and transmittance NIR measurements gave rise to better accuracy than that based on the diffuse-reflectance measurement,and the best PLS regression was obtained with external validation set having coefficient of determination R^2=0.97 and root mean square errors of prediction(RMSEP)=0.27%.For the assessment of protein content in single rice grains,it is still hard to establish one quantitative model under diffuse-reflectance measurement because of the effect of rice husk.The quantitative model based on the transflectance NIR spectral data yielded a relatively higher accurate result(RMSEP=0.81%),while the optimal result was obtained under the transmittance measurement resulting in R^2=0.96 and RMSEP=0.24%.This work may provide a practical solution for rapid screening of single rice seeds.