Study On Quantitative Structure-activity Relationship By Novel Molecular Structural Characterization

Molecular structural characterization (MSC) has become quite indispensable to drug design and pharmacodynamic evaluations, meanwhile the first step involved in modern quantitative structure–activity relationship (QSAR) studies. Encoding information on topological or steric characteristics of molecular structure, MSC dedicates to minimize information loss during this process. Prevalent MSC methods nowadays mainly include two kinds: one is based on molecular two dimensional (2D) structures and the other on three dimensional (3D) configurations. Based on the research of predecessor, two novel vectorial structural descriptors, called atomic electronegativity interaction vector (AEIV) and three-dimensional holographic vector of atomic interaction field (3D-HoVAIF), are developed by our gourp. They are used to describe the structures of drugs especially organic ones based on the interactions, expressed by atomic electronegativity and distance between various atoms, and atomic types. And they have been analyzed initially. The main contents are as follows:①All of 854 equivalent resonance carbon atoms from 39 lignans of butane's derivatives are parameterized by atomic electronegativity interaction vector (AEIV) and atomic hybridization state index (AHSI), then successfully model 13CNMR chemical shifts of these 854 carbon atoms. The correlation coefficient value of QSSR model estimation based on multiple linear regression analysis is R2=0.962, the leave-one-out cross-validation (LOOCV) Q2=0.925. Afterwards, this model is tested by 20 13CNMR chemical shift of new lignans from pine needles of pinus massoniana Lamb, and the result is similar to that of experiment.The results show that the model with AEIV and AHSI is of satisfactory estimation stability and favorable generalization.②In order to further discuss the applicable fields of AEIV, it is also used to characterize the molecular structure of Mannich bases. Models are established between the structure and the infrared spectroscopy vibration frequency of the C=O and P=O bonds in Mannich bases. It shows that the correlation between AEIV and infrared spectroscopy vibration frequency of C=O and P=O bonds in Mannich bases is good.③A three-dimensional holographic vector of atomic interaction field (3D-HoVAIF) is used to describe the chemical structure of fatty acids in ranunculus ternatus thunb, aim to study the quantitative structure-activity relationship (QSAR) of 18 fatty acids in ranunculus ternatus thunb. After the structural characterization, the descriptors obtained are screened by stepwise multiple regression (SMR). Then a partial least square regression(PLS) model is built. The correlation coefficient (R2) of established model, Leave-One-Out (LOO) Cross-Validation (CV) correlation coefficient(Q2) are 0.977 and 0.946, respectively . It is shown that the model has favorable stability and good prediction capability and the 3D-HoVAIF is applicable to the molecular structural characterization and biological activity prediction.④To study the quantitative structure-activity relationship (QSAR) of 30 acylthiourea analogue, 3D-HoVAIF is employed to describe the chemical structures of acylthiourea analogue. After the structure parameterization, the descriptors obtained are screened by stepwise multiple regression (SMR). Then a partial least square regression(PLS) model is built. The correlation coefficient R2 of established model, Leave-One-Out (LOO) Cross-Validation (CV) correlation coefficient Q2 are 0.624 and 0.409, respectively. It's shown that hydrophobic interaction has most important effect on the activity of acylthiourea analogue; the model has favorable stability and good prediction capability and the 3D-HoVAIF is applicable to the molecular structural characterization and biologicalactivity prediction.⑤For the first time, 3D-HoVAIF is employed to predict electrophoretic mobilities of dipeptides in capillary zone electrophoresis (CZE). To explore the usefulness of empirical models and multivariate analysis techniques in predicting electrophoretic mobilities of dipeptides in CZE, a stepwise multiple regression-multiple linear regression (SMR-MLR) model has been built. The data set consists of electrophoretic mobilities, measured at pH 2.5, for 53 dipeptides. Among the existing empirical models, the Offord model (i.e.,μ=Q/M2/3) gives the best correlation for the data set. A quantitative structure-mobility relationship (QSMR) has been developed using the Offord's charge-over-mass term as one descriptor combined with the 3D-HoVAIF descriptors to account for the steric, hydrophobic and electrostatic interactions of the amino acid side chains. The MLR results of the data set shows an improvement in the predictive ability of the model over the simple Offord's relationship.