A Combination Of Chemometrics And GC-MS Techniques In Application Of Edible Oil And Tumor Cell Classification

Gas chromatography-mass spectrometry is a highly efficient separation and analysis techniques. This technique use gas chromatography separation capacity to make separation of components in a mixture, and mass spectrometric identification of components for qualitative and quantitative analysis. Because of a very high sensitivity and relatively broad range of analysis, it’s becoming an indispensable component of the laboratory analytical instruments. As one of the main instruments and analytical tools, it has been widely used in food safety, environmental protection, disease prevention and control, chemical products, earth science, forensic science, military science and other fields.Appling chemometrics methods in food and biological of some congeners in similar properties and constructures with complex multicomponent spectral structure what’s more, exploring different ways to solve practical problems and the level of application of the model. There are important theoretical significance and guiding role in enriching the connotation of chemometrics and widening its scope of application. The main achievements of this thesis are as follows:1)Different composition and content of fatty acids in edible oil has certain differences in oil quality testing, which is used as indicators to identify the purity of fatty acids in edible oil information. Each oil has its characteristic fatty acid content of vegetable oil, mass spectrometry analysis requires the specimen pure or comparatively pure, all components should be characterized by noninterference mass spectra. Gas chromatography of fatty acids in edible oil complex can be separated and fingerprint analysis of fatty acids obtained, but gas chromatography of itself does not have a reliable qualitative ability. We can use the chromatographic separation of substances into the mass spectrometer to get the characteristic of the fatty acids in edible oil to solve this problem. Therefore, the use of gas chromatography-mass spectrometry analysis of fatty acids in edible oil that qualitative results and quantitative results can be obtained at the same time. Further more, in this thesis, oil methyl esterification method doesn’t need to heat, easily operate and saving much time. Because of complex contents and overlapping peaks of edible oil, or different kinds of edible oils having the spectrum too similar to a direct distinction, chemometric method can resolve this classification problem. For improved and accurate classification of edible vegetable oils with respect to type, for the first time support vector machine (SVM) as optimized using particle swarm optimization (PSO) algorithm is employed to construct the classification model based on the fatty acid profiles that obtained using gas chromatography-mass spectrometry. The combination of PSO and SVM can enable a flexible modeling manner for SVM according to the performance of the total model. It also can make the modeling technique to be an adaptive parameter-free method for edible vegetable oil classification. Results of the classification of six different kinds of edible vegetable oils reveal that the proposed strategy is of great promise in rapid and accurate identification of edible vegetable oils.2) Cancer has become one of the major disease a serious threat to human life and health, to adopt effective methods of prevention and treatment of tumor, so restraining the growth of malignant tumor is imminent. Changes of membrane phospholipids by influencing the various physical and chemical properties of the membrane, thereby affecting various biological functions of membrane proteins. Many diseases including tumors is accompanied by changes of membrane phospholipid metabolism, study on changes of membrane phospholipid not only helps to clarify the pathogenesis of these diseases and tumors, but also contribute to the diagnosis and treatment of disease. GC-MS analysis of phospholipid fatty acid composition differences in tumor cells, however, peaks found in Gas chromatographic are often overlapped and hardly distinguished. In this thesis, different kinds of tumor cells categorized by partial least squares regression model, this model can achieve regression modeling, simplification of the correlation between the two variables and simplify data structures with a very good explanation of the variation of the response variable information, the calibration stability of the model predictions is robust. Tumor and many diseases are associated with cell membrane phospholipid metabolism changes, because the cell membrane phospholipids containing fatty acid chain, but fatty acid chain of different phospholipids are different, so we can get quantitative information of phospholipids by fatty acid methyl esterification, and then according to the fatty acid chain difference of different kinds of cancer cells by using chemometrics methods, which can accurately classify fatty acid profile data of cancer cells. Using gas chromatography-mass spectrometry technology analysis of tumor cell phospholipid fatty acid composition differences are considered in this paper, but gas chromatography-mass spectrometry technology often has certain limitations, such as overlapping gas chromatographic peak and different types of cancer cells. Therefore, we can apply chemometric methods to solve the classification problem. In this paper, we use partial least squares regression model to classify different kinds of tumor cells, this model is a method of correlation analysis between two group variables and data structure and can well explain the variability of the response variable. Pathogenesis through changes in membrane phospholipid fatty acid research not only helps to elucidate the tumor and many other diseases, which has important significance for the diagnosis and treatment of the disease.3)In order to meet the demand of tobacco industry modernization, the research in science, reasonable simulation of physical and biological process based on human smoking cigarettes, according to oral, tracheal and nasal absorption characteristics and the absorption liquid composition characteristics, we established the pretreatment and analysis of test technology of biomimetic absorption type, the flue gas of bionic nearly40different brand of the specifications of the cigarette absorption liquid chemical composition data. At the same time, with the cigarette smoking experience of experts, the sensory quality of the cigarette (such as artificial). Smoking is based on the analysis of the results of testing and expert, this study further combined with chemometric modeling method, design and construct different regression models, comprehensive performance effects of different types of models in tobacco sensory quality prediction, in-depth study of flue gas bionic absorption liquid chemical proteomics data to predict the feasibility of the sensory quality of cigarette on the, to obtaining the ideal model to predict the smoke biomimetic absorption liquid chemical proteomics data based on cigarette quality.