Computational Descriptors Research Articles

Developing 2D-QSAR models for naphthyridine derivatives against HIV-1 integrase activity

HIV-1 integrase is an extremely important nominee in developing new and effective drugs especially naphthyridine compounds against acquired immune deficiency syndrome. The quantitative structure–activity relationship (QSAR) modeling is the most powerful method in computer-aided drug design and will be used to help the design of new naphthyridine derivatives. Different computational 2D-QSAR procedures applied to predict the relationship between the computational descriptors of naphthyridine derivatives with their HIV-1 integrase inhibition activities. Four different models including stepwise-MLR, consensus stepwise-MLR, GAPLS-MLR, and consensus GAPLS-MLR with appropriate correlation between the calculated and experimental biological activities (pIC50) against HIV-1 integrase were generated. Predictive QSAR models were obtained with R training 2 values of 0.848, 0.862, 0.709, and 0.751 as well as R test 2 values of 0.521, 0.651, 0.502, and 0.775 for stepwise-MLR, consensus stepwise-MLR, GAPLS-MLR, and consensus GAPLS-MLR models, respectively. QSAR models are high efficiency in prediction of the pIC50 in comparison with other models because of concerning the combination of “quantum and molecular mechanical” descriptors. Combination of “quantum” and “molecular mechanical” descriptors improved our models with high efficient test set activity prediction potency. The obtained results provided useful information for understanding the effects of polarizability, electronegativity, and especially functional groups such as aromatic nitrogens that are important for the activities of naphthyridine compounds. The developed QSAR models will be efficient for the rational design of potent naphthyridine derivatives against HIV-1 integrase activity.

Quantitative analysis of molecular interaction potentials of ionic liquid anions using multi-functionalized stationary phases in HPLC.

The molecular interaction potentials, including S (dipolarity/polarizability), A (hydrogen bonding acidity), and B (hydrogen bonding basicity), of anions are experimentally determined using multi-functionalized stationary phases in high-performance liquid chromatography (HPLC) systems. We employ three different multi-functionalized stationary phase columns (Obelisc R, Obelisc N, and Acclaim Trinity-P1) combined with two ingredients, namely, acetonitrile (ACN) and methanol (MeOH). These conditions can cause neutral, cationic, and anionic compounds to be retained. By using the retention characteristics of calibration compounds, including cations, anions, and neutral compounds, system parameters including the ionic interaction terms (zc Zc , za Za ) are evaluated using multiple linear regression, resulting in a standard deviation (SD) of 0.090-0.158 log units. Based on the system parameters and retention characteristics of the anions of interest, their molecular interaction potentials are characterized on the same scale for neutral and cationic molecules. Furthermore, to verify the determined molecular interaction potentials, we predict anion hydrophobicity. The results show that the determined S, A, and B, together with the computable descriptors E (excess molar refraction) and V (McGowan volume), can predict anion hydrophobicity with R(2) =0.982 and SD=0.167 (dimensionless).

Oxidation-Potential Tuning of Tungsten–Alkylidyne Complexes over a 2 V Range

The electrochemistry and electronic structures of over 30 tungsten-alkylidyne compounds of the form W(CR)L(n)L'(4-n)X (R = H, Bu(t), Ph, p-C6H4CCH, p-C6H4CCSiPr(i)3; X = F, Cl, Br, I, OTf, Bu(n), CN, OSiMe3, OPh; L/L' = PMe3, 1/2 dmpe, 1/2 depe, 1/2 dppe, 1/2 tmeda, P(OMe)3, CO, CNBu(t), py), in which the alkylidyne R group and L and X ligands are systematically varied, have been investigated using cyclic voltammetry and density functional theory calculations in order to determine the extent to which the oxidation potential may be tuned and its dependence on the nature of the metal-ligand interactions. The first oxidation potentials are found to span a range of ∼2 V. Symmetry considerations and the electronic-structure calculations indicate that the highest occupied molecular orbital (and redox orbital) is of principal d(xy) orbital parentage for most of the compounds in this series. The dependence of the oxidation potential on ligand is a strong function of the symmetry relationship between the substituent and the d(xy) orbital, being much more sensitive to the nature of the equatorial L ligands (π symmetry, with respect to d(xy), ΔE1/2 ≅ 0.5 V/L) than to the axial CR and X ligands (nonbonding with respect to d(xy), ΔE(1/2) < 0.3 V/L). The oxidation potential is linearly correlated with the calculated d(xy) orbital energy (slope ≅ 1, R(2) = 0.97), which thus provides a convenient computational descriptor for the potential. The strength of the correlation and slope of unity are proposed to be manifestations of the small inner-sphere reorganization energy associated with one-electron oxidation.

Predicting Skin Sensitization Potency for Michael Acceptors in the LLNA Using Quantum Mechanics Calculations

This study outlines the development of a series of quantitative mechanistic models enabling skin sensitization potency in the LLNA to be predicted for direct acting Michael acceptors. These models utilized several computational descriptors based on knowledge of the Michael addition reaction mechanism. The key descriptor was calculated using density functional theory and modeled the stability of the reaction intermediate. A second descriptor relating to the available surface area at the site of the reaction was also found to be important. Several poorly predicted compounds were identified, and in all cases, these could be rationalized mechanistically. The analysis of these compounds allowed a well-defined mechanistically driven applicability domain to be developed. The study showed that in silico quantitative mechanistic models, with a well-defined applicability domain, can be used to predict skin sensitization potency in the LLNA. The approach presented has the potential to be of use as part of a weight of evidence approach for predicting skin sensitization without the use of animals in risk assessment.

Combined 3D-QSAR modeling and molecular docking study on multi-acting quinazoline derivatives as HER2 kinase inhibitors.

A series of new quinazoline derivatives has been recently reported as potent multi-acting histone deacetylase (HDAC), epidermal growth factor receptor (EGFR), and human epidermal growth factor receptor 2 (HER2) inhibitors. HER2 is one of the major targets for the treatment of breast cancer and other carcinomas. Three-dimensional structure-activity relationship (3D-QSAR) is a well-known technique, which is used to drug design and development. This technique is used for quantitatively predicting the interaction between a molecule and the active site of a specific target. For each 3D-QSAR study, a three-dimensional model is created from a large curve fit to find a fitting between computational descriptors and biological activity. This model could be used as a predictive tool in drug design. The best model has the highest correlation between theoretical and experimental data. Self-Organizing Molecular Field Analysis (SOMFA), a grid-based and alignment-dependent 3D-QSAR method, is employed to study the correlation between the molecular properties and HER2 inhibitory potency of the quinazoline derivatives. Before presentation of inhibitor structures to SOMFA study, conformation of inhibitors was determined by AutoDock4, HyperChem and AutoDock Vina, separately. Overall, six independent models were produced and evaluated by the statistical partial least square (PLS) analysis. Among the several generated 3D-QSARs, the best model was selected on the basis of its statistical significance and predictive potential. The model derived from the superposition of docked conformation with AutoDock Vina with reasonable cross-validated q(2) (0.767), non cross-validated r(2) (0.815) and F-test (97.22) values showed a desirable predictive capability. Analysis of SOMFA model could provide some useful information in the design of novel HER2 kinase inhibitors with better spectrum of activity.

Molecular Modeling Descriptors and Analytical Chemistry Data of Antihistaminic Drugs

The relationships between experimental and computational descriptors of antihistamine drugs were studied using principal component analysis (PCA). Empirical data came from UV and IR spectroscopic measurements. Nonempirical data, such as structural molecular descriptors and chromatographic data, were obtained from HyperChem software. Another objective was to test whether the parameters used as independent variables (nonempirical and empirical-spectroscopic) could lead to attaining classification similar to that developed on the basis of the chromatographic parameters. To arrive at the answer to the question, a matrix of 18x49 data, including HPLC and UV and IR spectroscopic data, together with molecular modeling studies, was evaluated by the PCA method. The obtained clusters of drugs were consistent with the drugs' chemical structure classification. Moreover, the PCA method applied to the HPLC retention data and structural descriptors allowed for classification of the drugs according to their pharmacological properties; hence it may potentially help limit the number of biological assays in the search for new drugs.

Exploring Quantitative Nanostructure-Activity Relationships (QNAR) Modeling as a Tool for Predicting Biological Effects of Manufactured Nanoparticles

Evaluation of desired and undesired, biological effects of Manufactured NanoParticles (MNPs) is of critical importance for the future of nanotechnology. Experimental studies, especially toxicological, are time-consuming and costly, calling for the development of efficient computational tools capable of predicting biological events caused by MNPs from their structure and physical chemical properties. This mini-review assesses the potential of modern cheminformatics methods such as Quantitative Structure - Activity Relationship modeling to develop statistically significant and externally predictive models that can accurately forecast biological effects of MNPs from the knowledge of their physical, chemical, and geometrical properties. We discuss major approaches for model building and validation using both experimental and computed properties of nanomaterials. We consider two different categories of MNP datasets: (i) those comprising MNPs with diverse metal cores and organic decorations, for which experimentally measured properties can be used as particle's descriptors, and (ii) those involving MNPs possessing the same core (e.g., carbon nanotubes), but different surface-modifying organic molecules, for which computational descriptors can be calculated for a single representative of the decorative molecule. We illustrate those concepts with three case studies for which we successfully built and validated predictive models. In summary, this mini-review demonstrates that, analogous to conventional applications of QSAR modeling for the analysis of datasets of bioactive organic molecules, its application to modeling MNPs that we term Quantitative Nanostructure Activity Relationship (QNAR) modeling can be useful for (i) predicting activity profiles of novel MNPs solely from their representative descriptors and (ii) designing and manufacturing safer nanomaterials with desired properties.

Computational Descriptors for Chelating P,P- and P,N-Donor Ligands1

The ligand knowledge base approach has been extended to capture the properties of 108 bidentate P,P- and P,N-donor ligands. This contribution describes the design of the ligand set and a range of DFT-calculated descriptors, capturing ligand properties in a variety of chemical environments. New challenges arising from ligand conformational flexibility and donor asymmetry are discussed, and descriptors are related to other parameters, such as the ligand bite angle. A novel map of bidentate ligand space, potentially useful in catalyst design and discovery, has been derived from principal component analysis of the resulting LKB-PP descriptors. In addition, a range of multiple linear regression models have been derived for both experimental and calculated data, considering ligand bite angles in square-planar palladium complexes and ligand dissociation energies from octahedral chromium complexes, respectively. These data sets were fitted with models based on LKB descriptors to explore the transferability of des...

Structural effects on the phosphorylation of 3-substituted 1-β- d-ribofuranosyl-1,2,4-triazoles by human adenosine kinase

The conversion of ribavirin to the monophosphate by adenosine kinase is the rate-limiting step in activation of this broad spectrum antiviral drug. Variation of the 3-substituents in a series of bioisosteric and homologated 1-β- d-ribofuranosyl-1,2,4-triazoles has marked effects on activity with the human adenosine kinase, and analysis of computational descriptors and binding models offers insight for the design of novel substrates.

Solubility Prediction of Drugs in Supercritical Carbon Dioxide Using Artificial Neural Network

The descriptors computed by HyperChem ® software were employed to represent the solubility of 40 drug molecules in supercritical carbon dioxide using an artificial neural network with the architecture of 15-4-1. The accuracy of the proposed method was evaluated by computing average of absolute error (AE) of calculated and experimental logarithm of solubilities. The AE (± SD) of data sets was 0.4 (±0.3) when all data points were used as training set and the solubilities were back-calculated. The AE for predicted solubilities using a trained network employing 1/3 of data points from each set was 0.4 (± 0.3) and this finding reveals that the network is well trained using a limited number of experimental data. To provide a full predictive method, data sets were divided into two sets and the network was trained using 20 data sets and the next 20 sets were used as prediction sets. The produced average AEs (± SD) were 1.7 (±1.1) and 1.6 (±1.5), for two sets of analyses. In these analyses, only the computational descriptors, temperature and pressure of SC-CO 2 were used and no experimental solubility data is employed.

Impact of EGFR point mutations on the sensitivity to gefitinib: Insights from comparative structural analyses and molecular dynamics simulations

Emergence of resistant mutations in drug targets represents a serious problem in the targeted chemotherapy. One challenging issue is to understand the atomic-detailed effect of the mutation on the target. Another intriguing issue is how to predict specific mutations that would show up in the clinical setting, leading to drug resistance. By computational approaches, we have investigated structural, dynamics and energetic effects of a series of EGFR mutations identified from the lung cancer patients. We demonstrated mutation L858R caused gefitinib move closer to the hinge region, whereas T790M caused the ligand escape from the binding pocket. In particular, the T790M decreased the size of the hydrophobic slot formed by L718 and G796. This suggests that, to be effective against the T790M mutant, the inhibitors should avoid interactions with the hydrophobic slot. Mutations T790M, L858R, and their combinations are found to cause different conformational redistribution and to perturb the electrostatic potential at the ATP-binding pocket. Normal mode analysis revealed the mutations resulted in changes in the correlated movements in the protein. In an attempt to develop a computational descriptor for predicting the functional effect of EGFR mutations, we have developed a Plarm algorithm, and the Plarm score was found to be an excellent predictor of the functional impact of six clinical relevant mutations in EGFR tyrosine kinase domains, including T790M, L858R, G719C, L861Q, T790M + L858R double mutant, and delL747-P753insS. The Plarm algorithm could be readily extended to investigate other drug targets.

Computer-aided diagnosis of emphysema in COPD patients: Neural-network-based analysis of lung shape in digital chest radiographs

Several abnormalities of the shape of lung fields (depression and flattening of the diaphragmatic contours, increased retrosternal space) are indicative of emphysema and can be accurately imaged by digital chest radiography. In this work, we aimed at developing computational descriptors of the shape of the lung silhouette able to capture the alterations associated with emphysema. We analyzed two-sided digital chest radiographs from a sample of 160 patients with chronic obstructive pulmonary disease (COPD), 60 of which were affected by emphysema, and from 160 subjects with normal lung function. Two different description schemes were considered: a first one based on lung-silhouette curvature features, and a second one based on a minimal-polyline approximation of the lung shape. Both descriptors were employed to recognize alterations of the lung shape using classifiers based on multilayer neural networks of the feed-forward type. Results indicate that pulmonary emphysema can be reliably diagnosed or excluded by using digital chest radiographs and a proper computational aid. Two-sided chest radiographs provide more accurate discrimination than single-view analysis. The minimal-polyline approximation provided significantly better results than those obtained from curvature-based features. Emphysema was detected, in the entire dataset, with an accuracy of about 90% (sensitivity 88%, specificity 90%) by using the minimal-polyline approximation.

A Fully Computational Model for Predicting Percutaneous Drug Absorption

The prediction of transdermal absorption for arbitrary penetrant structures has several important applications in the pharmaceutical industry. We propose a new data-driven, predictive model for skin permeability coefficients k(p) based on an ensemble model using k-nearest-neighbor models and ridge regression. The model was trained and validated with a newly assembled data set containing experimental data and structures for 110 compounds. On the basis of three purely computational descriptors (molecular weight, calculated octanol/water partition coefficient, and solvation free energy), we have developed a model allowing for the reliable, purely computational prediction of skin permeability coefficients. The model is both accurate and robust, as we showed in an extensive validation (correlation coefficient for leave-one-out cross validation: Q = 0.948, mean standard error: 0.2 for log k(p)).

Enhancing the hit-to-lead properties of lead optimization libraries

In this paper we address several issues in the design of lead optimization libraries. Multipharmacophore descriptors were first developed in the context of designing diverse compound libraries. One reason for favoring such descriptors is the importance of the pharmacophore hypothesis in understanding the interaction of a compound with a protein target. Allied to this is the proposal that sampling over all potential pharmacophores leads to diversity in a biologically relevant space. We present results in support of this argument and also demonstrate that such methods are applicable to the design of focused libraries where the aim is to design the library toward a known lead or leads. This portability is important because it means that the same descriptors can be used for diverse library design, screening set selection, and focused library design, giving a consistent approach. We also examine the question of designing libraries with improved pharmacokinetic properties and show that it is possible to derive simple and rapidly computable descriptors applicable to the prediction of drug transport properties. Furthermore, these can be applied in the context of library design, although it may be necessary to synthesize libraries in a noncombinatorial manner to obtain the best results. To address this problem, we describe a Monte Carlo search procedure that allows the selection of a near-combinatorial subset in which all library members satisfy the design criteria. We present an example from our own work that illustrates how consideration of calculated log P, molecular weight, and polar surface area in the design of a combinatorial library can lead to compounds with improved absorption characteristics as determined by experimental Caco-2 measurements.

Calculation of Octanol/Water Partition Coefficients for Pesticides: A Comparative Study

On the basis of a set of 593 experimental octanol/water partition coefficients (log P) for pesticides, the simulation performances of two models using computable descriptors are compared. The back propagation neural network model designed from autocorrelation descriptors (SAR QSAR Environ. Res. (1997), 7, 151–172) compares favorably with the AFC model (J. Pharm. Sci. (1995), 84, 83–92) using atom/fragment contributions and correction factors.

Multivariate analysis of experimental and computational descriptors of molecular lipophilicity.

Two experimental (log P, R(Mw)) and 17 calculation descriptors for molecular lipophilicity (fragmental, atom-based or based on molecular properties) were investigated by multvariate analysis for a database of 159 compounds including both simple structures as well as more complex drug molecules. Principal component analysis (PCA) of the entire database exhibits a clustering of chemical groups; preciseness of clustering corresponds to chemical similarity. Thus, diversity searching in databases might effectively be performed by PCA on the basis of calculated log P. The comparative validity check of experimental and computational procedures by regression analysis and PCA was performed with a chemically balanced, reduced data set (n = 55) representing 11 chemical groups with 5 members each. Regression of experimental descriptors (log Poct versus RMW) proves that chromatographic data, obtained under well-defined experimental conditions, can be used as valid substitutes for log P. Regression of calculated versus experimental lipophilicity data shows a superiority of fragmental over atom-based methods and approaches based on molecular properties, as indicated by correlation coefficients, slopes and intercepts. In addition, PCA revealed that fragmental methods (Rekker-type, KOWWIN, KLOGP) sense the compound ranking in log P data to almost the same extent as experimental approaches. For atom-based procedures and CLOGP, both the comparability of absolute values and the sensing of the compound ranking in the database are slightly less. This trend is more pronounced for the methods based on molecular properties, with the exception of BLOGP.

A finite element code for thin plate dynamics

The present paper offers a remarkably simple alternative dynamic finite element computational descriptor for thin plate vibrational capability. The spirit for the proposed displacement version finite element formalism rests upon the problem setup procedure of the recently published automated hybrid finite element system of the same author (A. and H. Alaylioglu, 1987). The major advantage of this simulator is the minimization of the element integration efforts, such that the number of integrand evaluations is as in the case of a commonly envisaged one-point Gauss integrated model, while the results are free from the hourglass instabilities. Thereby, the scheme circumvents the computational expense involved in a priori or a posteriori anti-hourglassing filter, that is, stabilization of the underintegrated element. The purpose of this paper is threefold: (i) to fully develop powerful algorithms pertaining to the computer automated finite element analysis system that is user-friendly, efficient and reliable, (ii) to apply them to a spectrum of practical vibration mode predictions, and (iii) to compare the solutions with alternative approaches. Two stringent benchmark test problems, consisting of a cantilever and a clamped plate, are used to demonstrate the efficacy of the proposed easy-to-use thin plate simulator based on nonreduced integral implementation.