基于近红外光谱-支持向量机技术识别单碱基差异

目的建立一种基于近红外光谱-支持向量机（SVM）技术的单碱基差异识别方法。方法以仅相差1个碱基对的4种双链DNA为研究对象,其近红外光谱为识别变量,以径向基核函数（RBF）SVM建立非线性识别模型。结果对于长度为100bp的DNA链,当正则化系数γ=0.1,惩罚系数C=106时,模型的支持向量数最小为32,识别正确率为100%。结论该方法可发展为一种新的检测单核苷酸多态性的方法,具有简单、快速、低成本等优点。

Identification of Single-base Variant by Using Near Infrared Spectrum-Support Vector Machines Technology

Objective To develop a novel approach based on near infrared spectrum-support vector machines technology for identifying single-base variant. Methods Four types of DNA strands with single-base variant were taken as subjects, and their near-infrared spectrums as identifying variables, the nonlinear recognition model for identifying the single-base variant were developed by using radial basis functions-support vector machines（RBF-SVM）. Results For 100-bp DNA strands, an identification rate of 100% with the minimal support vector number of 32 was obtained when the regularization coefficient γ was 0.1 and the penalty coefficient C was 106. Conclusion This method provides a promising new means and new ideas for genetic polymorphism detection with advantages of simplness, fastness, and inexpensiveness.