Xiang-Yuan Li

Continuous Medium Theory for Nonequilibrium Solvation: I. How to Correctly Evaluate Solvation Free Energy of Nonequilibrium

Considering the influences of electrostatic potential Φ upon the change of solute charge distribution δρ and ρ upon the change δΦ at the same time, a more reasonable integral formula of dG = (1/2) ∫V (ρδΦ + Φδρ)dV is used to calculate the change of the electrostatic free energy in charging the solute–solvent system to a nonequilibrium state, instead of the one of dG = ∫V ΦδρdV used before. This modification improves the expressions of electrostatic free energy and solvation free energy, in which no quantity of the intermediate equilibrium state is explicitly involved. Detailed investigation reveals that the solvation free energy of nonequilibrium only contains the interaction energy between the field due to the solute charge in vacuum, and the dielectric polarization at the nonequilibrium state. The solvent reorganization energies of forward and backward electron transfer reactions have been redefined because the derivations lead to a remarkable feature that these quantities are direction‐dependent, unlike the theoretical models developed before. The deductions are given in the electric field‐displacement form. Relevant discussions on the reliability of theoretical models suggested in this work have also been presented

Continuous Medium Theory for Nonequilibrium Solvation: II. Interaction Energy between Solute Charge and Reaction Field and Single-Sphere Model for Spectral Shift

On the basis of continuous medium theory, a model for evaluation of spectral shifts in solution has been developed in this work. The interaction energy between solute dipole and reaction field and the self‐energy of the reaction field have been formulated through derivations. Applying the interaction energy expression together with the point dipole approximation to the case of spherical cavity produces new formulations of spectral shifts. The same expression of electrostatic free energy of the nonequilibrium state is achieved by integrating the change of the electrostatic free energy for a charging process. Moreover, generalized formulations evaluating spectral shifts have been established in the charge‐potential notation, and the reduction of them to the point dipole case consistently leads to the same formulations of spectral shifts as those by interaction energy approach. Mutual supports provide convincing evidences for the reliability of the present results. In this work, attentions are particularly paid to the conclusion of zero self‐energy of the reaction field, which is different from the previous theory. Reasoning and arguments are given on this point. From the present derivations, it is concluded that the spectral shifts of light absorption and emission were theoretically exaggerated in the past, in particular, by a factor of 2 for the spectral shift sum.

Prediction of chemical carcinogenicity by machine learning approaches

In this paper we report a successful application of machine learning approaches to the prediction of chemical carcinogenicity. Two different approaches, namely a support vector machine (SVM) and artificial neural network (ANN), were evaluated for predicting chemical carcinogenicity from molecular structure descriptors. A diverse set of 844 compounds, including 600 carcinogenic (CG+) and 244 noncarcinogenic (CG−) molecules, was used to estimate the accuracies of these approaches. The database was divided into two sets: the model construction set and the independent test set. Relevant molecular descriptors were selected by a hybrid feature selection method combining Fischers score and Monte Carlo simulated annealing from a wide set of molecular descriptors, including physiochemical properties, constitutional, topological, and geometrical descriptors. The first model validation method was based a five-fold cross-validation method, in which the model construction set is split into five subsets. The five-fold cross-validation was used to select descriptors and optimise the model parameters by maximising the averaged overall accuracy. The final SVM model gave an averaged prediction accuracy of 90.7% for CG+ compounds, 81.6% for CG− compounds and 88.1% for the overall accuracy, while the corresponding ANN model provided an averaged prediction accuracy of 86.1% for CG+ compounds, 79.3% for CG− compounds and 84.2% for the overall accuracy. These results indicate that the hybrid feature selection method is very efficient and the selected descriptors are truly relevant to the carcinogenicity of compounds. Another model validation method, i.e. a hold-out method, was used to build the classification model using the selected descriptors and the optimised model parameters, in which the whole model construction set was used to build the classification model and the independent test set was used to test the predictive ability of the model. The SVM model gave a prediction accuracy of 87.6% for CG+ compounds, 79.1% for CG− compounds and 85.0% for the overall accuracy. The ANN model gave a prediction accuracy of 85.6% for CG+ compounds, 79.1% for CG− compounds and 83.6% for the overall accuracy. The results indicate that the built models are potentially useful for facilitating the prediction of chemical carcinogenicity of untested compounds.

Identification of small molecule aggregators from large compound libraries by support vector machines

Small molecule aggregators non‐specifically inhibit multiple unrelated proteins, rendering them therapeutically useless. They frequently appear as false hits and thus need to be eliminated in high‐throughput screening campaigns. Computational methods have been explored for identifying aggregators, which have not been tested in screening large compound libraries. We used 1319 aggregators and 128,325 non‐aggregators to develop a support vector machines (SVM) aggregator identification model, which was tested by four methods. The first is five fold cross‐validation, which showed comparable aggregator and significantly improved non‐aggregator identification rates against earlier studies. The second is the independent test of 17 aggregators discovered independently from the training aggregators, 71% of which were correctly identified. The third is retrospective screening of 13M PUBCHEM and 168K MDDR compounds, which predicted 97.9% and 98.7% of the PUBCHEM and MDDR compounds as non‐aggregators. The fourth is retrospective screening of 5527 MDDR compounds similar to the known aggregators, 1.14% of which were predicted as aggregators. SVM showed slightly better overall performance against two other machine learning methods based on five fold cross‐validation studies of the same settings. Molecular features of aggregation, extracted by a feature selection method, are consistent with published profiles. SVM showed substantial capability in identifying aggregators from large libraries at low false‐hit rates.

Continuous medium theory for nonequilibrium solvation: III. Solvation shift by monopole approximation and multipole expansion in spherical cavity

According to the classical electrodynamics, a new and reasonable method about electrostatic energy decomposition of the solute‐solvent system has been proposed in this work by introducing the concept of spring energy. This decomposition in equilibrium solvation gives the clear comprehension for different parts of total electrostatic free energy. Logically extending this cognition to nonequilibrium leads to the new formula of electrostatic free energy of nonequilibrium state. Furthermore, the general solvation shift for light absorption/emission has been reformulated and applied to the ideal sphere case with the monopole approximation and multipole expansion. Solvation shifts in vertical ionizations of atomic ions of some series of main group elements have been investigated with monopole approximation, and the variation tendency of the solvation shift versus atomic number has been discussed. Moreover, the solvation shift in photoionization of nitrate anion in glycol has been investigated by the multipole expansion method.

Theoretical investigation of electron transfer transition in tetracyanoethylene-contained organic complexes

In this work, the authors use complete active space self‐consistent field method to investigate the photoinduced charge‐separated states and the electron transfer transition in complexes ethylene‐tetracyanoethylene and tetramethylethylene‐tetracyanoethylene. Geometries of isolated tetracyanoethylene, ethylene, and tetramethylethylene have been optimized. The ground state and the low‐lying excited states of ethylene and tetracyanoethylene have been optimized. The state energies in the gas phase have been obtained and compared with the experimentally observed values. The torsion barrier of tetracyanoethylene has been investigated through the state energy calculation at different conformations. Attention has been particularly paid to the charge‐separated states and the electron transfer transition of complexes. The stacked conformations of the donor–acceptor complexes have been chosen for the optimization of the ground and low‐lying excited states. Equilibrium solvation has been considered by means of conductor‐like screening model both in water and in dichloromethane. It has been found that the donor and tetracyanoethylene remain neutral in complexes in ground state 1A1 and in lowest triplet state 3B1, but charge separation appears in excited singlet state 1B1. Through the correction of nonequilibrium solvation energy based on the spherical cavity approximation, π→π* electron transfer transition energies have been obtained. Compared with the experimental measurements in dichloromethane, the theoretical results in the same solvent are found higher by about 0.5 eV.

A theoretical study of solvent effect and reaction mechanism of electron transfer reaction between indole side chain and phenol side chain of peptide involving tryptophan and tyrosine

The mechanisms for the electron transfer between the indole side chain and phenol side chain of peptides involving tryptophan and tyrosine have been studied by ab initio calculation in this work. The solvent effect has been considered by using the conductor‐like screening model (COSMO). Three mechanisms including the two‐proton elimination mechanism, the hole transfer without proton deprotonation, and the step‐by‐step mechanism for the electron transfer processes between indole and phenol have been discussed. From the theoretical calculations, we can conclude that, for the electron transfer between indole and phenol, and consequently between tryptophan and tyrosine, a deprotonation process is necessary, otherwise the ET process is difficult to proceed. There are two deprotonation cases that make the successive electron transfer feasible: one is that both the proton attached to indole N and that attached to phenol O, are removed before the ET process, the other is that the proton attached to indole N is first removed, then the proton attached to O of phenol group transfers to the indole group, and the ET follows. The comparison between our theoretical predication and the experimental conclusion has been made. The influence of the donor‐acceptor distance on the ET rate constant for the intramolecular electron transfer has been investigated in solution.

Accurate prediction of enthalpies of formation for a large set of organic compounds

This article describes a multiparameter calibration model, which improves the accuracy of density functional theory (DFT) for the prediction of standard enthalpies of formation for a large set of organic compounds. The model applies atom based, bond based, electronic, and radical environmental correction terms to calibrate the calculated enthalpies of formation at B3LYP/6‐31G(d,p) level by a least‐square method. A diverse data set of 771 closed‐shell compounds and radicals is used to train the model. The leave‐one‐out cross validation squared correlation coefficient q2 of 0.84 and squared correlation coefficient r2 of 0.86 for the final model are obtained. The meanabsolute error in enthalpies of formation for the dataset is reduced from 4.9 kcal/mol before calibration to 2.1 kcal/mol after calibration. Five‐fold cross validation is also used to estimate the performance of the calibration model and similar results are obtained.

Prediction of Antitumor Activity for Epothilone Analogues Based on 3D Molecular Descriptors

Abstract In order to predict the antitumor activities of various epothilone analogues, a set of molecular descriptors, including electronic, topological, and geometrical descriptors and molecular shape indices ( K -order moment shape indices), were calculated to characterize the structural and physicochemical properties for 150 compounds. The 30 descriptors selected with genetic algorithm were employed to establish the classification model of epothilone analogues by using support vector machine (SVM). This SVM system gives a total prediction accuracy of 83.3% by the ‘leave-one-out’ (LOO) method and that of 80.6% by the five-fold cross-validation method. The present study indicates that the K -order moment shape indices defined by us are useful for the description of configuration isomers, and SVM is a facilitating tool in prediction of the antitumor activity of epothilone analogues.