Applications Of QSAR In Life Analytical Chemistry And Environmental Chemistry

Quantitative structure-activity relationship (QSAR) methods, are the most promising and successful tools to provide rapid and useful meanings for predicting the biological activity or toxicities of organic compounds by use of different statistical methods and various kinds of molecular descriptors. The goal of QSAR is to develop models on a training set of compounds, these models will then allow for the prediction of the biological activity of related chemicals. This kind of study can not only develop a method for the prediction of the property of compounds that have not been synthesized but also can identify and describe important structural features of molecules that are relevant to variations in molecular properties, thus gain some insight into structural factors affecting molecular properties. Now, QSAR method has been introduced to biology, chemistry and environment science. In this dissertation, we mainly discussed a novel machine learning method, support vector machine (SVM) to construct QSAR model.Chapter 1 of the dissertation included a brief description of the QSAR history, principle, realization process and research status. In this section, we also introduced the support vector machine method and a review of the application of SVM in biology, chemistry and environment area.Chapter 2 described the application of QSAR method to analytical chemistry and biology. A brief introduction was given as follows: (1) Two classification models, linear discriminant analysis (LDA) and SVM, were developed to construct a reliable QSAR model to predict/classify milk-to-plasma (M/P) drug concentration ratio and to distinguish the potential risk of drugs to nursing infants. The classification accuracy of training set. and test set for SVM was 90,63 and 90.00%, respectively. The total accuracy for SVM was 90.48%, which was higher than that of LDA (77.78%). Comparison of the two methods shows that the performance of SVM was better than that of LDA, which implies that the SVM method is an effective tool in evaluating the risk of drugs when experimental M/P ratios have not been investigated. (2) SVM algorithm was used to develop QSAR model for predicting the binding affinity of 152 nonapeptides, which can bind to class I MHC HLA-A*201 molecule. The heuristic method (HM) was used to select proper descriptors responsible for binding affinity and build linear model. The best results are found using SVM: RMS errors for training, test and whole data set were 0.371, 0.382 and 0.377, respectively. This paperallow the prediction of the binding affinity of new, untested peptides and, through the analysis of contribution of each parameter of different residue at specific position of peptidic ligands, to understand nature of the forces governing binding behavior and suggest new ideas for further synthesis of high-affinity peptides. (3) MLR, RBFNN and SVM methods were applied to predict folding rate of 28 two-state proteins. The results demonstrate that SVM model is the best and native-state topology is the major determinant for the folding rates of two-state proteins. (4) MLR and SVM algorithm was used to build linear and non-linear models to predict ligand-protein binding affinities. RMS errors of the two methods for test were 1.57 and 0.79, respectively. (5) SVM method was used to develop a predictive model for early diagnosis of anorexia. It was based on the concentration of six elements (Zn, Fe, Mg, Cu, Ca, and Mn) and the age extracted from 90 cases. Compared with the results obtained from two other classifiers, PLS and BPNN, the SVM method exhibited the best whole performance. The accuracies for the test set by PLS, BPNN, and SVM methods were 52%, 65%, and 87%, respectively. Moreover, the models we proposed could also provide some insight into what factors were related to anorexia.Chapter 3 described the application of QSAR method to environmental science. Chemical byproducts from industrial systems that are allowed to escape into the environment can have toxic effects. Each of these chemicals has the potential to be harmful, and it is crucial that each compound be assessed for its toxicity level. There are several experimental methods available for screening the toxicity of chemicals. However, this can be costly, time-consuming, and could potentially pro-duce toxic side products from the experimental methods used today. This has meant that the development of computational methods as an alternative tool for predicting properties of chemicals has been a subject of intensive study. QSAR method has been proved to be one of the most promising and successful methods not only to provide rapid and useful meanings of predicting toxicity of chemicals but also to aid understanding of the mechanism. A brief introduction was given as follows: (1) QSAR models were constructed using MLR, RBFNN and SVM methods to predict the relative binding affinity (expressed as log (RBA)) to the androgen receptor (AR) of 146 endocrine disrupting compounds (EDCs). Comparison of the results obtained from three models showed that the SVM method exhibited the best overall performances. The RMS error of MLR, RBFNN and SVM models for test set were 0.96, 1.03 and 0.59. (2) MLR, RBFNN and SVM methods were applied for predicting 76 compounds' toxicity,which were expressed as negative logarithm of the effective concentration (log(l/EC50)), at which microtoxs reduce the emission of bioluminescence by 50%. The RMS error of MLR, RBFNN and SVM models for test set were 0.75, 0.51 and 0.35. It showed that the SVM is a powerful tool to predict toxicity of chemicals. (3) The polybrominated diphenyl ethers (PBDEs) are a group of aromatic brominated compounds used in abundance as additive flame-retardants in polymers found in textiles and electrical appliances. They are emerging as a significant class of environmental contaminants and the pollution of PBDEs has become increasing in recent years. In this work, 2D and 3D QSAR methods were applied to study the toxicity of PBDEs. In addition, linear models using HM was built to predict two important properties, vapor pressure (VP) and octanol-air partition coefficient (KOA), of PBDEs. It can help us to accurately asses the risk of PBDEs to environment and human health.