In an effort to facilitate drug discovery, computational methods for facilitating the prediction of various adverse drug reactions (ADRs) have been developed. So far, attention has not been sufficiently paid to the development of methods for the prediction of serious ADRs that occur less frequently. Some of these ADRs, such as torsade de pointes (TdP), are important issues in the approval of drugs for certain diseases. Thus there is a need to develop tools for facilitating the prediction of these ADRs. This work explores the use of a statistical learning method, support vector machine (SVM), for TdP prediction. TdP involves multiple mechanisms and SVM is a method suitable for such a problem. Our SVM classification system used a set of linear solvation energy relationship (LSER) descriptors and was optimized by leave-oneout cross validation procedure. Its prediction accuracy was evaluated by using an independent set of agents and by comparison with results obtained from other commonly used classification methods using the same dataset and optimization procedure. The accuracies for the SVM prediction of TdP-causing agents and non-TdP-causing agents are 97.4 and 84.6% respectively; one is substantially improved against and the other is comparable to the results obtained by other classification methods useful for multiple-mechanism prediction problems. This indicates the potential of SVM in facilitating the prediction of TdP-causing risk of small molecules and perhaps other ADRs that involve multiple mechanisms.

Analysis of vibrational motions and thermal fluctuational dynamics is a widely used approach for studying structural, dynamic and functional properties of proteins and nucleic acids. Development of a freely accessible web server for computation of vibrational and thermal fluctuational dynamics of biomolecules is thus useful for facilitating the relevant studies. We have developed a computer program for computing vibrational normal modes and thermal fluctuational properties of proteins and nucleic acids and applied it in several studies. In our program, vibrational normal modes are computed by using modified AMBER molecular mechanics force fields, and thermal fluctuational properties are computed by means of a selfconsistent harmonic approximation method. A web version of our program, MoViES (Molecular Vibrations Evaluation Server), was set up to facilitate the use of our program to study vibrational dynamics of proteins and nucleic acids. This software was tested on selected proteins, which show that the computed normal modes and thermal fluctuational bond disruption probabilities are consistent with experimental findings and other normal mode computations. MoViES can be accessed at http://ang.cz3.nus. edu.sg/cgi-bin/prog/norm.pl.

Ab initio molecular orbital method and Monte Carlo (MC) simulation with free energy perturbation (FEP) techniques have been used to study the aza-Wittig reaction of iminophosphoranes (H3PdNH) with formaldehyde (H2CO) in the gas phase and in three different solvents: water, methanol, and tetrahydrofuran (THF). The optimized structures and thermodynamic properties of stationary points for the title reaction system in the gas phase were calculated at the MP2/6-31G** level of theory. The effects of substituents on the reactivity of iminophosphorane were discussed. This aza-Wittig reaction is more favorable for XdH and CH3 than for XdCl in the gas phase. The potential energy profiles along the minimum energy path in the gas phase and in three solvents were obtained. The solvent effects on the H3PdNH + H2CO reaction increase in the order water ≈ methanol>THF, suggesting that the protic polar solvents are more suitable for the aza-Wittig reaction.

The aza-Wittig reaction of iminophosphoranes (X3PdNH, X=H and Cl) with formaldehyde (H2CO) was investigated in gas phase and in water using ab initio MP2/6-31G** level of theory and the self-consistent reaction field theory (isodensity polarized continuum model, IPCM). In the gas phase, the aza-Wittig reaction was predicted to be a two-step process with two dipole-dipole complexes, one four-membered ring intermediate and two transition states. The potential energy profiles along the minima energy path in gas phase and in water were obtained. The solvent effects on the thermodynamic and kinetic properties of this reaction were discussed. This aza-Wittig reaction is more favorable for X=H than for X=Cl, both in the gas phase and in water

The chemical fixation of nitric oxide (NO) reacting with alkynyllithium to produce 5-methyl-3-oxide-1,2,3-oxadiazole has been investigated by using ab initio (U)MP2 and DFT/(U)B3LYP methods. The solvent effect was assessed using the combination of microsolvation model with explicit THF ligands on lithium and continuum solvent model based on the SCRF/CPCM method at the (U)-B3LYP/6-31G* level. Our results reveal that the overall reaction is stepwise and considered to include two processes. In process 1, the nitrogen atom in nitric oxide at first attacks the C1 atom in alkynyllithium to afford the intermediate 5. In process 2, after another nitric oxide reacted with the intermediate 5 to produce 8a, we found that two pathways are involved. For path 1, the O2 atom at first attacks the C2 atom to form a five-membered ring geometry, and then lithium can rotate around the N1-O1 bond, leading to the product 5-methyl-3-oxide-1,2,3-oxadiazole followed addition of water. However, for path 2, lithium atom rotates first around the N1-O1 bond, and then the product 5-methyl-3-oxide-1,2,3-oxadiazole is also generated by addition of water. Our calculations indicate that path 1 is more favorable than path 2 in the gas phase, while both of them exist possibly in THF solvent. The overall reaction is exothermic.

P-glycoproteins (P-gp) actively transport a wide variety of chemicals out of cells and function as drug efflux pumps that mediate multidrug resistance and limit the efficacy of many drugs. Methods for facilitatingearly elimination of potential P-gp substrates are useful for facilitating new drug discovery. A computational ensemble pharmacophore model has recently been used for the prediction of P-gp substrates with a promising accuracy of 63%. It is desirable to extend the prediction range beyond compounds covered by the known pharmacophore models. For such a purpose, a machine learning method, support vector machine (SVM), was explored for the prediction of P-gp substrates. A set of 201 chemical compounds, including 116 substrates and 85 nonsubstrates of P-gp, was used to train and test a SVM classification system. This SVM system gave a prediction accuracy of at least 81.2% for P-gp substrates based on two different evaluation methods, which is substantially improved against that obtained from the multiple-pharmacophore model. The prediction accuracy for nonsubstrates of P-gp is 79.2% using 5-fold cross-validation. These accuracies are slightly better than those obtained from other statistical classification methods, including k-nearest neighbor (k-NN), probabilistic neural networks (PNN), and C4.5 decision tree, that use the same sets of data and molecular descriptors. Our study indicates the potential of SVM in facilitating the prediction of P-gp substrates.

Statistical-learning methods have been developed for facilitating the prediction of pharmacokinetic and toxicological properties of chemical agents. These methods employ a variety of molecular descriptors to characterize structural and physicochemical properties of molecules. Some of these descriptors are specifically designed for the study of a particular type of properties or agents, and their use for other properties or agents might generate noise and affect the prediction accuracy of a statistical learning system. This work examines to what extent the reduction of this noise can improve the prediction accuracy of a statistical learning system. A feature selection method, recursive feature elimination (RFE), is used to automatically select molecular descriptors for support vector machines (SVM) prediction of P-glycoprotein substrates (P-gp), human intestinal absorption of molecules (HIA), and agents that cause torsades de pointes (TdP), a rare but serious side effect. RFE significantly reduces the number of descriptors for each of these properties thereby increasing the computational speed for their classification. The SVM prediction accuracies of P-gp and HIA are substantially increased and that of TdP remains unchanged by RFE. These prediction accuracies are comparable to those of earlier studies derived from a selective set of descriptors. Our study suggests that molecular feature selection is useful for improving the speed and, in some cases, the accuracy of statistical learning methods for the prediction of pharmacokinetic and toxicological properties of chemical agents.

*: PM3), and ONIOM (MP2/6-31G*: HF/3-21G)-were applied to investigate thermal decomposition mechanisms of four 2-phenoxycarboxylic acids (2-phenoxyacetic acid, 2-phenoxypropionic acid, 2-phenoxybutyric acid, and 2-phenoxyisobutyric acid) in the gas phase. All the transition states and intermediates of the reaction paths were optimized. The reaction pathway of four reactants yielding the phenol, CO, and the corresponding carbonyl compound was characterized on the potential energy surface and found to proceed stepwise. The first step corresponds to the elimination of phenol and the formation of α-lactone intermediate through a five-membered ring transition state, and the second step is the cycloreversion process of α-lactone intermediate to form CO and the corresponding carbonyl compound. The reaction pathway of latter three compounds to produce the carboxylic acid and phenol via a four-membered cyclic transition structure was also examined theoretically. Comparison with experiment indicates that the activation parameters for the fist reaction channel are accurately predicted at the ONIOM (MP2/6-31G*: HF/3-21G) level of theory.

Density functional theory (DFT) and Monte Carlo (MC) simulation with free energy perturbation (FEP) techniques have been used to study the tautomeric proton transfer reaction of 2-amino-2-oxazoline, 2-amino-2-thiazoline, and 2-amino-2-imidazoline in the gas phase and in water. Two reaction pathways were considered: the direct and water-assisted transfers. The optimized structures and thermodynamic properties of stationary points for the title reaction system in the gas phase were calculated at the B3LYP/6-311 G(d, p) level of theory. The potential energy profiles along the minimum energy path in the gas phase and in water were obtained. The study of the solvent effect of water on the proton transfer of 2-amino-2-oxozoline, 2-amino-2-thiazoline, and 2-amino-2-imidazoline indicates that water as a solvent is favorable for the water-assisted process and slows down the rate of the direct transfer pathway.

Toxicity of various compounds has been measured in many studies by their toxic effects against Tetrahymena pyriformis. Efforts have also been made to use computational quantitative structure-activity relationship (QSAR) and statistical learning methods (SLMs) for predicting Tetrahymena pyriformis toxicity (TPT) at impressive accuracies. Because of the diversity of compounds and toxicity mechanisms, it is desirable to explore additional methods and to examine if these methods are applicable to more diverse sets of compounds. We tested several SLMs (logistic regression, C4.5 decision tree, k-nearest neighbor, probabilistic neural network, support vector machines) for their capability in predicting TPT by using 1129 compounds (841 TPT and 288 non-TPT agents) which are more diverse than those in other studies. A feature selection method was used for improving prediction performance and selecting molecular descriptors responsible for distinguishing TPT and non-TPT agents. The prediction accuracies are 86.9% 94.2% for TPT and 71.2% 87.5% for non-TPT agents based on 5-fold cross-validation studies, which are comparable to some of earlier studies despite the use of more diverse sets of compounds. The selected molecular descriptors are consistent with those used in other studies and experimental findings. These suggest that SLMs are useful for predicting TPT potential of diverse sets of compounds and for characterizing the molecular descriptors associated with TPT.