应用近红外光谱技术实现转双价基因(Cry1Ab/Cry2Aj-G10evo)玉米的快速识别  

Discrimination of Transgenic Maize Containing the Cry1Ab/Cry2Aj and G10evo Genes Using Near Infrared Spectroscopy(NIR)

作　　者：彭城 冯旭萍 何勇[2] 张初[2] 赵懿滢 徐俊锋[1] PENG Cheng;FENG Xu-ping;HE Yong;ZHANG Chu;ZHAO Yi-ying;XU Jun-feng(State Key Laboratory Breeding Base for Zhejiang Sustainable Pest andDisease Control,Zhejiang Academy of Agricultural Sciences,Hangzhou　310021,China;College of Biosystems Engineering and Food Science,Zhejiang University,Hangzhou　310058,China)

机构地区：[1]浙江省农业科学院,浙江省植物有害生物防控重点实验室-省部共建国家重点实验室培育基地,浙江杭州310021 [2]浙江大学生物系统工程与食品科学学院,浙江杭州310058

出　　处：《光谱学与光谱分析》2018年第4期1095-1100,共6页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(31471417,31501280),浙江省自然科学基金项目(LQ15C130002)和国家高技术研究发展计划(863)项目(2013AA102301)资助

摘　　要：转基因技术在过去的几十年里快速发展,然而此项技术对生态环境、伦理道德等可能带来的影响尚存争议,因此针对农作物的转基因成分检测和鉴别的相关技术研究十分重要。本研究以转双价基因(cry1Ab/cry2Aj-G10evo)玉米籽粒和玉米面粉为研究对象,采用近红外光谱仪采集900~1 700nm波段范围的光谱,结合Savitzky-Golay(SG)平滑算法对提取出的光谱数据进行去除噪声处理。基于全波段光谱和PCA主成分分别建立了偏最小二乘判别分析(PLS)和支持向量机判别模型(SVM)。试验结果表明,在转基因玉米籽粒全谱的判别分析模型中,SVM判别模型效果要优于PLS判别模型,SVM模型识别正确率达到90%以上,PLS的模型识别率只有85%左右。以PCA降维后建立的模型中,SVM模型也取得了最优的效果,建模集和预测集识别正确率达到100%。虽然转基因玉米在研磨加工后外源蛋白和DNA有所下降,但是转基因玉米粉末基于全波段光谱建立的SVM模型的建模集正确率仍有90.625%。结果表明应用近红外光谱技术集合化学计量学方法对转基因玉米的鉴别是可行的,为转基因玉米乃至其他转基因农产品的鉴别提供了技术支持,具有重要的理论意义和应用价值。Genetic engineering technique has made rapid strides in the past decades,however,the potential problems of this technique for environmental,ethical and religious impact are unknown.It is necessary to research on the detection of genetically modified organisms in agricultural crops and in products derived.In the present study,Near infrared spectroscopy(NIR)combined with chemometrics was successfully proposed to identify transgenic and non-transgenic maize.Transgenic maize single kernel and flour containing both cry1Ab/cry2Aj-G10evo protein and their parent,non-transgenic ones were measured in NIR diffuse reflectance mode with spectral range of 900~1 700 nm.Savitzky-Golay(SG)was used topreprocess the selection spectral region with absolute noises.Two classification methods,partial least square(PLS)and support vector machine(SVM):were used to build discrimination models based on the preprocessed full spectra and the dimension reduction information extracted by principal component analysis(PCA).Discriminant results of transgenic maize kernel based on SVM obtained a better performance by using the preprocessed full spectra compared to PLS model.The SVM achieved more than 90%calibration accuracy,while the PLS obtained just about 85%accuracy.By applying the PCA dimension reduction of theNIRreflectance in conjunction with the SVM model,the discrimination of transgenic from non-transgenic maize kernel was with accuracy up to 100%for both calibration set and validation set.The correct classification for transgenic and non-transgenic maize flour was 90.625%using SVM based on preprocessed full spectra,although degration of exogenous gene and protein existed during the milling.The results indicated that INR spectroscopy techniques and chemometricsmethods could be feasible ways to differentiate transgenic maize and other transgenic food.

关 键 词：近红外光谱 转双价基因玉米 偏最小二乘判别分析模型 支持向量机判别模型 

分 类 号：S123[农业科学—农业基础科学]