Abstract Discrimination of genetically modified organisms is increasingly required by legislation and consumers worldwide. Currently the most commonly used detection methods for identification of transgenic crops are high cost, destructive and time-consuming, so not suitable for fast and extensive application. Raman is a noninvasive and nondestructive spectroscopic technique capable of extracting sample fingerprints. In this paper, Raman spectroscopy and chemometric tools were evaluated for discrimination of transgenic corn. Different spectral preprocessing as well as algorithms for variables selection were evaluated to fit a classifier model based on linear discriminant analysis (LDA). Results showed spectral bands assigned to carbohydrates and carotenoids responsible for classes discrimination. The best classifier achieved 87.5 % of predictive accuracy. These results suggest that genetic differences between evaluated classes are also expressed in their chemical composition, which could be detected using Raman spectroscopy. The developed method is clean, fast and can contribute for establishing normative about genetically modified foods.

中文翻译：

---

用于区分转基因玉米的拉曼光谱

摘要 全世界的立法和消费者越来越需要对转基因生物进行歧视。目前最常用的转基因作物鉴定检测方法成本高、破坏性强、耗时长，不适合快速广泛应用。拉曼是​​一种能够提取样品指纹的非侵入性和非破坏性光谱技术。在本文中，拉曼光谱和化学计量学工具对转基因玉米的鉴别进行了评估。评估了不同的光谱预处理以及变量选择算法，以适应基于线性判别分析 (LDA) 的分类器模型。结果显示分配给负责分类的碳水化合物和类胡萝卜素的光谱带。最好的分类器达到了 87。5% 的预测准确度。这些结果表明，评估类别之间的遗传差异也体现在它们的化学成分上，这可以使用拉曼光谱进行检测。开发的方法干净、快速，有助于建立转基因食品的规范。