QSPR study on Hydrophobicity of Pt(II) complexes with surface electrostatic potential-based descriptors.

Pt(II) complexes play an important role in bioinorganic chemistry due to their antitumor activities. In the present study, we focused on building predictive models for the hydrophobicity of Pt(II) complexes. A five-parameter model, integrating frontier orbital energies (EHOMO, ELUMO) and descriptors derived from electrostatic potentials on molecular surface, was firstly constructed by using multiple linear regression (MLR) method. Mechanistic interpretations of the introduced descriptors were elucidated in terms of intermolecular interactions in the n-octanol/water partition system. Then, four up-to-date modeling methods, including support vector machine (SVM), least-squares support vector machine (LSSVM), random forest (RF) and Gaussian process (GP), were utilized to build the nonlinear models. Systematical validations including leave-one-out cross-validation, the validation for test set, as well as a very rigorous Monte Carlo cross-validation (MCCV) were performed to verify the reliability of the constructed models. The peak, median and integralRext2 values of the best GP model are 0.88, 0.86 and 0.84, respectively. The root mean squared errors for the test set (RMSEP) of the MLR, SVM, LSSVM and GP models fall in the range of 0.62-0.71. Although they are not superior to prior models built with the use of a number of descriptors, the results are satisfactory. Applicability domain of the model was evaluated.

Prediction of the hydrophobicity of platinum(IV) complexes based on molecular surface properties.

A quantitative structure-property relationship (QSPR) study was performed for predicting the hydrophobicity of Pt(IV) complexes. Two four-parameter equations, one based solely on structural descriptors derived from electrostatic potentials (ESPs) on molecular surface, and the other integrated ESP descriptors with molecular surface area (AS), were firstly constructed. Mechanistic interpretations of the structural descriptors introduced were elucidated in terms of solute-solvent intermolecular interactions. Subsequently, several up-to-date modeling techniques, including support vector machine (SVM), least-squares support vector machine (LSSVM), random forest (RF) and Gaussian process (GP), were utilized to build the nonlinear models. Systematical validations including leave-one-out cross-validation, the validation for test set, as well as a more rigorous Monte Carlo cross-validation were performed to verify the reliability of the constructed models. The predictive performances of the four different nonlinear modeling methods follow the order of LSSVM≈GP > RF > SVM. The pure-ESP-based models are generally inferior to the AS-integrated ones. Comparisons with previous results were made.