Studies On Genotyping Of STR And SNP With Near Infrared Spectroscopy And Chemical Pattern Recognition

Based on near infrared spectroscopy (NIRS), the paper has done some studies on a simple, rapid and low-cost detection method of short tandem repeats (STRs) and single nucleoide polymorphisms (SNPs).To establish a typing method of short tandem repeats (STRs) based on NIRS-chemical pattern recognition. Taking the three different genotypes 10-10, 10-11 and 11-11 of locus D16S539 as example, which have a small degree of difference, DNA fragments containing the polymorphism sites were amplified by a pair of primers to obtain three genotypes samples; these samples were tested by the NIRS directly; using their spectra as recognition variables, first, selecting support vector machine (SVM) which showed a better ability for both linear and nonlinear problem and owned prominent advantages for modeling ability of small samples to do the initial exploration experiment, the SVM model of the three genotypes was established, and had good stability for a few calibration samples, the prediction rate of 100 %, the mean squared error of 0, and the squared correlation coefficient of 1.00. At the same time, choosing a simple linear multivariate calibration method the principal discriminant variate (PDV) set up the three genotypes discriminant model, and the result of classification was visual and had a good separability and stability, the accuracy of prediction samples was up to100%.In order to practical application, collecting STR samples with high gene frequency (i.e. 9-9, 9-11 and 11-11), the conditions of PCR amplification and NIRS detection for a DNA fragment containing the polymorphism sites were studied, and then the standardized amplification and detection conditions have been established, under which standard samples and standard NIRS-s were obtained. The linear PDV discriminant model of STRs with high gene frequency was established, and owned good separability and stability, the prediction rate of 100 %. And further increased the sample numbers, at the same time, added a class of samples(i.e.9-12), four kinds of genotype samples were classified by the PDV method, and the discrimination tree structure of multiclass STRs was preliminarily established. The established STRs discriminant model had a good fitting, stability, and strong prediction (i.e. the predicting accuracy was 100%) by adjusting the parametersλ, which was a preliminary basic research for building expert system of multiclass STRs typing.For studying the detection of SNP genotype, taking a SNP (857G>A) of N-acetyltransferase 2 (NAT2) as an example, DNA fragments containing the SNP site were amplified by the PCR method based on a pair of primers to obtain the three-genotype (GG, AA, GA) modeling samples. Based on the NIRS of the measured samples and back propagation artificial neural network (BP-ANN), the genotype discriminant model of SNP has been established. The compressed variables of the NIRS-s as the discriminant ones, the rich information of the compressed variables was extracted based on the other ANN. For the established discriminant model, the root mean square error for the training and the prediction sample sets were 0.0135 and 0.0132, respectively. The accuracy of prediction samples was up to100%.Based on NIRS and chemical pattern recognition techniques, the genotypes of STR and SNP could be determined by one step PCR amplification and detection of NIRS without any preprocessing for the analyzed samples after PCR, which was simple, rapid and low-cost.