The Quantitative Analysis Of Biological Molecules Based Upon SERS Spectra And The Atomic Spectral Classification By Chemical Pattern Recognition

The nucleic acid and protein are two basic composition of thebiological substance, especially for the nucleic acid, as basic geneticmaterial, it plays an important role in the protein synthesis process.Therefore the research work about nucleic acid should be paid muchattention. Because the nucleic acid is macromolecular material, thestudy is relatively complex. Qualitative and quantitative analysisresearch with basic groups, as the main composition of nucleic acid,will help to understand the physiological activities and biologicalcharacteristics. Such research had been restricted due to their relativecontent in vivo is low. When biological molecules are adsorbed uponcoarse metal surface or sol particles, the Raman scattering signalintensity of biological molecules will be significantly increased, and thisphenomenon is called surface-enhanced Raman scattering (SERS). It issignificant that SERS technique can be used for qualitative andquantitative analysis at very low molecular concentration level. SERStechnique is mainly dependent on SERS substrate and the substrate hasa direct and great impact on the strength of the SERS signals. Theroughened metallic surface, the aggregate and the colloidal sol arethe most generally used SERS substrates.In this article, the silver sol was used as SERS substrate for SERSquantitative analysis. Having considered the fact that silver sol is lessstable, polyacrylic acid sodium were added into the silver sol toimprove its stability. As for the mixture of adenine and guanine, theirSERS characteristic peaks overlap to considerable extent, and thisrestricted the application of conventional quantitative analysismethods.Chemometrics not only utilizes the chemical methods but alsoemploys the basic theory and method from the statistics, mathematics and computer science and so on to obtained chemical informationhidden in analytical data. Thus it becomes a powerful and useful tool toexplore chemical information, especially when we deal with a complexmixture, multivariate calibration algorithm in chemometrics plays agreat role. In this paper we try to use chemometrics techniques toanalyze the SERS spectra from the mixture of adenine and guanine forthe purpose of quantitative analysis.In addition, chemical pattern recognition techniques were alsoused to deal with the atomic spectral classification problem.Chemometrics methods also have the huge potential and advantagesas to the chemical pattern recognition. As for the even-parity andodd-parity spectral classification of Cm II, as long as the four features:energy level, the total angular momentum quantum number (J),Langde factor (g) and isotope shift (IS) are known, its electronicconfiguration is determined exclusively. According to the existingexperimental data, only those samples whose four features and theircorresponding configuration are known can be treated as knownsamples and used to establish a pattern classification model. If fourfeatures are known but its electronic configuration is unknown, it will betaken as an unknown sample. Through chemometrics methods trainingwith known samples, a pattern classification model can be establishedand used to predict the configuration assignment of an unknownsample. In this paper, chemometrics techniques based on BPN andSVM are used to solve the spectral classification of Cm II atomicspectroscopic data. It is shown that chemometrics can play a positiveand effective role in such classification study.This paper is composed by the following three aspects:(1) Quantitative analysis of adenine or guanine by SERSspectroscopy.(2) Application of chemometrics methods for SERS spectroscopic quantitative analysis with the mixtures of adenine and guanine.(3) Applications of the support vector machine and the error backpropagation neural networks for Cm II spectral data classification.