STR Genotyping Based On Spectroscopy And Chemical Patterm Recognization

By using spectra of short tandem repeat (STR) genotype samples as recognitionvariables, this paper had done some studies on genotyping based on chemical patternrecognition, to propose a simple, rapid and low-cost method for STR detection.A detection method based on near infrared spectra (NIRS)-principal discriminantvariate (PDV) was proposed for STR genotyping. Firstly, three genotypes10-10,10-11and11-11of D5S818locus with a middle length of the core-repeat units of nucleotidesAGAT and the high frequencies were selected as the research subjects for methodologyresearch. According to the optimum polymerase chain reaction (PCR), samples used forestablishing models were obtained. Using near-infrared spectra of samples as therecognition variables, the optimal PDV model for these three genotypes was built. Theobtained model had good separability, stability and discriminating power. Subsequently,in order to know whether this method has the feasibility for classification of thegenotypes more than three, this paper selected six genotypes of D5S818locus toestablish the expert system. According to the difference of total number of core tandemrepeat, the five PDV models in the expert system all had satisfactory separability androbustness, which demonstrated that NIRS-PDV could successful implementation forSTR genotyping.In order to make the STR genotyping more simply, a detection method based onultraviolet spectra (UVS) was studied. Firstly, the UVS based PDV model for genptypes10-11,11-11and11-12was established. The model had good separability andpredictive ability, three studied genotypes all could classified correctly. Nextly, basedon the difference of total number of core tandem repeat, an expert system for sixgenotypes of D5S818was constructed. In this expert system, the five PDV models inthe expert system all had good separability and robustness, six studied genotypes allcould discriminated correctly, which indicated that UVS-PDV could be used for STRgenotyping. Simultaneously, a UVS based partial least squares discriminant (DPLS) was also used for detection genotypes10-11,11-11and11-12, the optimal DPLS model alsohad good discrimination for three studied genotypes.In the last part of this paper, D16S539locus which is widely applied in forensicmedicine was selected as research objective to investigate the applicability of theUVS-PDV method. Firstly, the genotypes whose difference was4core tandem repeatswere selected to establish the model, and they could be discriminated correctly. Then,the genotypes with the same total number of core repeat units were also discriminatedrightly according to the afore-method. The results indicated that the PDV could be usedto discriminate the genotypes with different degrees of difference. Eventually, sixgenotypes were utilized to study the expert system based on their different total numberof core repeat units, and they were discriminated rightly. The results further proved thatUVS-PDV could be successfully used for STR genotyping.Based on spectroscopy and chemical pattern recognition technology, STR couldachieve genotyping by one step PCR amplification and detection of spectra without anypreprocessing for the analyzed samples after PCR, which is simple, rapid and low-cost.