【摘要】 利用紫外光谱结合支持向量机(SVM)模式识别原理建立了短串联重复序列(STR)的分型方法。以D5S818基因座的10-10、11-13和13-13三个基因型为研究对象,优化了包含该多态性位点的DNA片段的聚合酶链反应(PCR)的扩增条件和紫外光谱检测条件,获得了优化条件下的待测样本紫外光谱。以该光谱为识别变量,建立了该3类基因型的SVM判别模型,有效地克服了光谱共线性,模型预测率100%。该方法无需任何检测前处理,只需一步PCR扩增和紫外光谱检测即可实现STR的基因分型,具有简单、快速、低成本等优点。更多还原

【Abstract】 An approach for genotyping of STR locus based on ultraviolet(UV) spectroscopy and support vector machine(SVM) was studied.Taking three different genotypes 10-10,11-13 and 13-13 of locus D5S818 as example,the conditions for PCR amplification and UV spectral detection on a DNA fragment containing the polymorphism sites were optimized.Under the optimal conditions,the ultraviolet spectra for the samples were obtained.Using these spectra as identifying variables,the SVM model of the three genotypes was established to overcome effectively the spectral colinearity,and a right prediction rate of 100% was obtained.The proposed method was simple,rapid and low-cost,and could be indirectly used to determine the three genotypes of D5S818 locus based on PCR amplification and UV spectra of the samples without any pretreatment.更多还原