Quantitative Structure-activity Relationship Descriptors Research Articles

Optimization of Antibacterial Peptides by Genetic Algorithms and Cheminformatics

Pathogens resistant to available drug therapies are a pressing global health problem. Short, cationic peptides represent a novel class of agents that have lower rates of drug resistance than derivatives of current antibiotics. Previously, we created a software system utilizing artificial neural networks that were trained on quantitative structure-activity relationship descriptors calculated for a total of 1400 synthetic peptides for which antibacterial activity was determined. Using the trained system, we correctly identified additional peptides with activity of 94% accuracy; active peptides were 47 of the top rated 50 peptides chosen from an in silico library of nearly 100,000 sequences. Here, we report a method of generating candidate peptide sequences using the heuristic evolutionary programming method of genetic algorithms (GA), which provided a large (19-fold) improvement in identification of novel antibacterial peptides. Approximately 0.50% of peptides evaluated during the GA method were classified as highly active, while only 0.026% of the nearly 100,000 sequences we previously screened were classified as highly active. A selection of these peptides was tested in vitro and activities reported here. While GA significantly improves the possibility of identifying candidate peptides, we encountered important pitfalls to this method that should be considered when using GA.

Docking and QSAR studies of β-phenylethylidenehydrazine derivatives as a Gamma-aminobutyric acid aminotransferase inhibitor

β-Phenylethylidenehydrazine (PEH) derivatives have been recognized as Gamma-aminobutyric acid aminotransferase (GABA-AT) inhibitors. In this research a group of newly synthesized of PEH analogs, possessing a variety of substituents (Me, OMe, Cl, and CF3) at the 2-, 3-, and 4-position of the phenyl ring, were subjected to docking study and quantitative structure activity relationship (QSAR) analysis. PEH analogs were built by HYPERCHEM program, and conformational studies were performed through semi-empirical method followed by PM3 method. QSAR descriptors were obtained from the EDRAGON and HYPERCHEM, and equations were derived from multilinear regression (MLR) method. The sums of the JGI2, H6m, and E2s were identified as the most significant descriptors. This simple equation can be used to estimate the GABA-AT inhibitory activity for new derivatives of this series of compounds. Docking study was performed by using AutoDock4 program on the all compounds. The obtained results show that the phenyl ring is inserted into the lipophilic pocket and that the NHNH2 moiety is situated in a mainly polar region of the enzyme. These computational studies can offer some useful references for understanding the action mechanism and performing the molecular design or modification of this series of GABA-AT inhibitors.

Local Indices for Similarity Analysis (LISA)—A 3D-QSAR Formalism Based on Local Molecular Similarity

A simple quantitative structure activity relationship (QSAR) approach termed local indices for similarity analysis (LISA) has been developed. In this technique, the global molecular similarity is broken up as local similarity at each grid point surrounding the molecules and is used as a QSAR descriptor. In this way, a view of the molecular sites permitting favorable and rational changes to enhance activity is obtained. The local similarity index, calculated on the basis of Petke's formula, segregates the regions into "equivalent", "favored similar", and "disfavored similar" (alternatively "favored dissimilar") potentials with respect to a reference molecule in the data set. The method has been tested on three large and diverse data sets-thrombin, glycogen phosphorylase b, and thermolysin inhibitors. The QSAR models derived using genetic algorithm incorporated partial least square analysis statistics are found to be comparable to the ones obtained by the standard three-dimensional (3D)-QSAR methods, such as comparative molecular field analysis and comparative molecular similarity indices analysis. The graphical interpretation of the LISA models is straightforward, and the outcome of the models corroborates well with literature data. The LISA models give insight into the binding mechanisms of the ligand with the enzyme and allow fine-tuning of the molecules at the local level to improve their activity.

Predicting highly-connected hubs in protein interaction networks by QSAR and biological data descriptors.

Hub proteins (those engaged in most physical interactions in a protein interaction network (PIN) have recently gained much research interest due to their essential role in mediating cellular processes and their potential therapeutic value. It is straightforward to identify hubs if the underlying PIN is experimentally determined; however, theoretical hub prediction remains a very challenging task, as physicochemical properties that differentiate hubs from less connected proteins remain mostly uncharacterized. To adequately distinguish hubs from non-hub proteins we have utilized over 1300 protein descriptors, some of which represent QSAR (quantitative structure-activity relationship) parameters, and some reflect sequence-derived characteristics of proteins including domain composition and functional annotations. Those protein descriptors, together with available protein interaction data have been processed by a machine learning method (boosting trees) and resulted in the development of hub classifiers that are capable of predicting highly interacting proteins for four model organisms: Escherichia coli, Saccharomyces cerevisiae, Drosophila melanogaster and Homo sapiens. More importantly, through the analyses of the most relevant protein descriptors, we are able to demonstrate that hub proteins not only share certain common physicochemical and structural characteristics that make them different from non-hub counterparts, but they also exhibit species-specific characteristics that should be taken into account when analyzing different PINs. The developed prediction models can be used for determining highly interacting proteins in the four studied species to assist future proteomics experiments and PIN analyses. THE SOURCE CODE AND EXECUTABLE PROGRAM OF THE HUB CLASSIFIER ARE AVAILABLE FOR DOWNLOAD AT: http://www.cnbi2.ca/hub-analysis/

Evaluating Different Descriptors for Model Design of Antimicrobial Peptides with Enhanced Activity Toward P. aeruginosa

The number of isolated drug-resistant pathogenic microbes has increased drastically over the past decades, demonstrating an urgent need for new therapeutic interventions. Antimicrobial peptides have for a long time been looked upon as an interesting template for drug optimization. However, the process of optimizing peptide antimicrobial activity and specificity, using large peptide libraries is both tedious and expensive. Here, we describe the construction of a mathematical model for prediction, prior to synthesis, of peptide antibacterial activity toward Pseudomonas aeruginosa. By use of novel descriptors quantifying the contact energy between neighboring amino acids in addition to a set of inductive and conventional quantitative structure-activity relationship descriptors, we are able to model the peptides antibacterial activity. Cross-correlation and optimization of the implemented descriptor values have enabled us to build a model (Bac2a- #2) that was able to correctly predict the activity of 84% of the tested peptides, within a twofold deviation window of the corresponding IC50 values, measured earlier. The predictive power, is an average of 10 submodels, each predicting the activity of 20 randomly excluded peptides, with a predictive success of 16.7 +/- 1.6 peptides. The model has also been proven significantly more accurate than a simpler model (Bac2a- #1), where the inductive and conventional quantitative structure-activity relationship descriptors were excluded.

QSAR − How Good Is It in Practice? Comparison of Descriptor Sets on an Unbiased Cross Section of Corporate Data Sets

The quality of QSAR (Quantitative Structure-Activity Relationships) predictions depends on a large number of factors including the descriptor set, the statistical method, and the data sets used. Here we study the quality of QSAR predictions mainly as a function of the data set and descriptor type using partial least squares as the statistical modeling method. The study makes use of the fact that we have access to a large number of data sets and to a variety of different QSAR descriptors. The main conclusions are that the quality of the predictions depends both on the data set and the descriptor used. The quality of the predictions correlates positively with the size of the data set and the range of biological activities. There is no clear dependence of the quality of the predictions on the complexity of the data set. All of the descriptors tested produced useful predictions for some of the data sets. None of the descriptors is best for all data sets; it is therefore necessary to test in each individual case, which descriptor produces the best model. In our tests, 2D fragment based descriptors usually performed better than simpler descriptors based on augmented atom types. Possible reasons for these observations are discussed.

Inductive Descriptors: 10 Successful Years in QSAR

The paper overviews the developments of 'A New Model of Inductive Effect' - an approach introduced 10 years ago for calculation of Taft's substituent constants. The original model enabled accurate quantification of inductive parameters σ * and allowed approaching numerous important theoretical problems associated with inductive and steric interactions. A number of methods derived from the original approach have been reviewed and discussed including those for 'inductive' electronegativity, 'inductive' hardness-softness and 'inductive' partial charges. The practical use of 'inductive' reactivity indices as a novel and effective class of QSAR (quantitative structure-activity relationships) descriptors has been illustrated in the context of QSAR studies of antibacterial activity of organic chemicals and cationic peptides. The further developments and prospective applications of 'inductive' 3D QSAR descriptors in the area of computer-aided drug design have also been discussed.

Ligand intramolecular motions in ligand-protein interaction: ALPHA, a novel dynamic descriptor and a QSAR study with extended steroid benchmark dataset.

The role of intramolecular motions in ligand-macromolecule interactions has been explored by developing and validating ALPHA, a novel QSAR (quantitative structure-activity relationship) descriptor. It is based on the spectral exponents (alpha), which measure the degree of 1/f alpha noise of coordinate fluctuations in molecular dynamics (MD) simulations. ALPHA is the first truly 'dynamic' QSAR descriptor, i.e., it can be derived directly from an MD trajectory. The performance of ALPHA was tested in detail employing the CBG (corticosteroid binding globulin) affinity of 31 benchmark steroids, supplemented with 11 steroids as an external test set. The only fair (42-50%) correlations of ALPHA with static 3D and electronic descriptors mean that ALPHA forms an independent molecular property. Furthermore, inclusion of ALPHA in the SOMFA/ESP model improves the correlation coefficient from 0.86 to 0.91, and /delta/ave from 0.46 to 0.36 for the benchmark dataset. The predictive ability of ALPHA can be interpreted as indirect evidence of the dynamic contribution to ligand-macromolecule interactions. The physical background of ALPHA is discussed and the importance of molecular motions for biological activity is anticipated.

Structure-activity relationships for the toxicity of polychlorinated dibenzofurans: approach through density functional theory-based descriptors.

The applicability of various density functional theory (DFT)-based descriptors--chemical softness, electronegativity, and electrophilicity index--for quantitative structure--activity relationships (QSARs) was investigated for polychlorinated dibenzofurans (PCDFs). The DFT descriptors were obtained by using the three parameter hybrid density functional, B3LYP, with the 6-311G(d,p) basis set. QSARs were developed relating aryl hydrocarbon receptor (AhR) binding affinities, aryl hydrocarbon hydroxylase and ethoxyresorufin O-deethylase induction potencies of PCDFs with DFT descriptors, hydrophobicity, and steric parameters. These QSARs explain around 75% of variation in AhR binding affinities of PCDFs. Congeners with higher toxicity values had larger softness values. Studies also showed that the most toxic isomer of tetrachlorodibenzofurans (TCDFs) and dibenzo-p-dioxins (TCDDs), respectively, had the largest chemical softness value in its respective group. The results show that DFT descriptors could be used as useful electronic descriptors in QSARs for the prediction of toxicity of PCDFs. Overall, 85 congeners of PCDFs and TCDDs were considered in this study.

Development of quantitative structure-activity relationships for the toxicity of aromatic compounds to Tetrahymena pyriformis: comparative assessment of the methodologies.

The purpose of this study was to develop quantitative structure-activity relationships (QSARs) for the toxicity of 268 aromatic compounds in the Tetrahymena pyriformis growth inhibition assay. The QSARs were developed using the response-surface (or two-parameter) approach, which was also modified using linear free-energy parameters to account for outliers. Subsequently, the data set was analyzed using partial least-squares (PLS). The results of the modeling using different methodologies were compared to the use of a Bayesian regularized neural network (BRANN) trained on the same data. Both response surface approaches, and PLS explained between 75 and 80% of the variance in the data; BRANN gave a higher statistical fit. In terms of the transparency of the approaches, the response surface clearly provides the simplest and easiest to use QSAR, it is readily interpreted in terms of mechanism of toxic action. PLS and BRANN are respectively less transparent. The use of atomistic and fragment-based indexes as descriptors in QSARs is assessed also, these are found not to be as useful as whole molecule parameters for the prediction of toxicity for molecules outside of the training set. The relative merits of the different approaches to the development of QSARs are described.

Structure–boiling point relationships of alkanes using the multifunctional autocorrelation method

The concept of the multifunctional autocorrelation method (MAM), governing the global description of molecules, has been changed to take into account the structural environment of each atom in all fragments of the molecule. New topological parameters are generated as possible descriptors in QSAR (quantitative structure–activity relationships) and can be useful for database characterization and encoding a variety of physicochemical properties. The well-known component Pk was modified as shown in the following formula: Pk=Σ[f(i)(Σ (f(kij))f(j)]1/n (n is the number of atoms in the fragment considered). The efficiency of the approach is demonstrated through a case study dealing with the design of a model structure–property relationship for boiling points of alkanes. The data set was analysed by multiple linear regression.

Antimycobacterial pyrroles: synthesis, anti-Mycobacterium tuberculosis activity and QSAR studies

A number of known antifungal pyrrole derivatives and some newly synthesized compounds (5–33) were tested in vitro against Mycobacterium tuberculosis CIP 103471. The majority of tested compounds were efficient antimycobacterial agents showing MIC values ranging from 0.5 to 32 μg/mL. A 3-D-QSAR study has been performed on these pyrrole derivatives to correlate their chemical structures with their observed inhibiting activity against M. tuberculosis. Due to the absence of information on a putative receptor responsible for this activity, classical quantitative structure–activity relationships (QSAR) and comparative molecular field analysis (CoMFA) have been applied. A model able to well correlate the antimycobacterial activity with the chemical structures of pyrrole derivatives 5–33 has been developed which is potentially helpful in the design of novel and more potent antituberculosis agents. The combination of CoMFA with classical QSAR descriptors led to a better hybrid 3-D-QSAR model, that successfully explains the structure–activity relationships (r2=0.86) of the training set. A comparison between the QSAR, CoMFA and mixed QSAR–CoMFA models is also presented. The hybrid model is to be preferred, however, because of its lowest values of the average absolute error of prediction toward a limited external test set.

Synthesis, pharmacological and biophysical characterization, and membrane-interaction QSAR analysis of cationic amphiphilic model compounds.

Cationic amphiphilic drugs have a propensity to interact with biological interphases. This study was designed to gain more insight into the molecular properties of catamphiphilic drugs which govern this type of interaction. A series of phenylpropylamine model compounds were synthesized in which modifications were incorporated at the aromatic part of the molecule. The replacement of (45)Ca(2+) from phosphatidylserine monolayers served to monitor drug binding to the phospholipid. The influence on the phase-transition temperature of liposomes of dipalmitoylphosphatidic acid was measured to assess the perturbing action of the drugs on the structural organization of phospholipid assemblies. The antiarrhythmic activity of the compounds was determined in Langendorff preparations of guinea pig hearts to assess the membrane-stabilizing action. Quantitative structure-activity relationship (QSAR) models for these endpoints were developed using both intra- and intermolecular QSAR descriptors. Intermolecular membrane-interaction descriptors were derived from molecular dynamics simulations of the compounds in a model phospholipid monolayer. QSAR models were derived for all endpoints using partial least-squares regression (PLS) and a genetic algorithm tool, the genetic function approximation (GFA). Membrane-interaction descriptors appear to be of a particular importance in explaining the influence of the compounds on the phase-transition temperature of DPPA liposomes, while the other endpoints can be adequately modeled by intramolecular descriptors. The calcium-displacing activity at phosphatidylserine monolayers is governed by the electrostatic properties of the compounds. Measures of lipophilicity and molecular size are of particular importance for antiarrhythmic activity. Possible improvements to both the molecular modeling and the applied computational protocol of membrane-solute systems are identified and discussed.

Quantitative structure-activity relationships (QSARs) of N-terminus fragments of NK1 tachykinin antagonists: a comparison of classical QSARs and three-dimensional QSARs from similarity matrices.

The ability of three-dimensional quantitative structure-activity relationships (QSARs) derived from classical QSAR descriptors and similarity indices to rationalize the activity of 28 N-terminus fragments of tachykinin NK1 receptor antagonists was examined. Two different types of analyses, partial least squares and multiple regression, were performed in order to check the robustness of each derived model. The models derived using classical QSAR descriptors lacked accurate quantitative and predictive abilities to describe the nature of the receptor-inhibitor interaction. However models derived using 3D QSAR descriptors based on similarity indices were both robust and significantly predictive. The best model was obtained through the statistical analysis of molecular field similarity indices (n = 28, r2 = 0.846, r(cv)2 = 0.737, s = 0.987, PRESS = 7.102) suggesting that electronic and size-related properties are the most relevant in explaining the affinity data of the training set. The overall quality and predictive ability of the models applied to the test set appear to be very high, since the predicted affinities of three test compounds agree with the experimentally determined affinities obtained subsequently within the experimental error of the binding data.

Using theoretical descriptors in quantitative structure-activity relationships: HPLC capacity factors for energetic materials

The widespread application of computational techniques to studies in biology and chemistry has led to a quest for important characteristic properties that may be directly derived from these methods. The theoretical linear solvation energy relationships (TLSER) have successfully replaced empirical parameters, such as in the linear solvation energy relationships (LSER) of Kamlet and Taft, with theoretical descriptors that are consistently derived for a wide variety of chemical and biological properties and for a large range of molecules. These descriptors are small in number, well-defined in theoretical terms, and provide a systematic basis for computational studies of complex phenomena. Energetic materials are a class of molecules whose explosive characteristics have been previously studied by empirical methods. The forensic and environmental sciences have prompted the conduct of detailed high performance liquid chromatography (HPLC) studies for identifying of trace quantities of commonly used explosives. The TLSER methodology provides good correlations between the calculated quantities and the experimentally determined HPLC capacity factors. This provides a reasonable interpretation of retention times in terms of molecular volume and quantities associated with acidity and basicity.

Frontier orbital energies in quantitative structure-activity relationships: A comparison of quantum chemical methods

The energies of the highest occupied and lowest unoccupied molecular orbitals (HOMO and LUMO) have long been used as descriptors in QSAR (Quantitative Structure-Activity Relationships). It is shown that different quantum chemical methods of calculating these energies yield results which sometimes correlate poorly with each other. This could seriously affect physical interpretation of QSAR equations. A comparison is made between HOMO and LUMO energies and their differences and sums (hardness and electronegativity) calculated by some of the best knownab initio and semi-empirical methods for two series of simple organic molecules. The difference between the HOMO and LUMO energies correlates better between methods than does either alone, and their sum correlates relatively poorly. MINDO/3 (Modified Intermediate Neglect of Differential Overlap, version 3) is the poorest method in terms of correlation with the more extended basis setab initio methods, followed by CNDO (Complete Neglect of Differential Overlap) and INDO (Intermediate Neglect of Differential Overlap). The best semi-empirical methods, in terms of correlation with experiment and the more extended basis setab initio calculations, are MNDO (Modified Neglect of Differential Overlap), AM1 (Austin Model 1) and PM3 (Parametric Method 3). The simplestab initio method, STO-3G, does not agree as well with the extended basis set calculations or with experimental results as the more advanced semi-empirical methods.

Rational estimation of the QSAR (quantitative structure-activity relationships) descriptors sigmaS0, and their applications for medicinals now available.

Rational estimation of the descriptor sigmaSo, representing the dispersion and repulsion energies in the van der Waals interaction for both aliphatic and aromatic moieties, enabled us to present the descriptors of several important medicinals now available. In this work, the fundamental rule for the estimation of the descriptor for a substrate having a variety of binding modes and the correction value deltaSo necessary for aliphatic heterocycle formation are confirmed, and the descriptors for several important moieties are established according to the concept of quantitative structure-activity relationship analogy. Furthermore, several kinds of herbicides, antiinflammatory agents, hypocholesterolemics, analgesics, sympathetic stimulants, and antipsychotics are concerned in this work.

Using theoretical descriptors in quantitative structure-activity relationships: some toxicological indices

The application of computational techniques to medicinal chemistry is growing at a tremendous rate. Quantitative structure-activity relationships (QSAR), which relate biological and toxicological activities to structural features, have been employed widely to correlate structure to activity. A difficulty of this approach has been nonuniformity of parameter sets and the inability to examine contributions across properties and data sets. Linear solvation energy relationships (LSER) developed by Kamlet and Taft circumvent many of the difficulties and successfully utilize a single set of parameters for a wide range of physical, chemical, and biological properties. We have replaced the LSER solvato-chromatic parameters with theoretically determined parameters to permit better a priori prediction of properties. Comparison of the two parameter sets for five biological activities is presented, showing the excellent fit of the theoretically determined parameters.

Estimation of the quantitative structure-activity relationship descriptor .SIGMA.S.DEG. for di- and tri-substituted benzene derivatives.

The quantitative structure-activity relationship descriptor sigma S0, representing the contributions from dispersion and repulsion interactions can be expressed as follows: (1) for disubstituted benzene derivatives, sigma S0(1, 2) = 0.831 sigma sigma S0(mono) - 0.004, sigma S0(1, 3) = 0.855 sigma sigma S0(mono) - 0.007, and sigma S0(1, 4) = 0.874 sigma sigma S0(mono) - 0.018; (2) for trisubstituted benzene derivatives, sigma S0(1, 3, 5) = 0.752 sigma sigma S0(mono) - 0.017, sigma S0(1, 2, 4) = 0.715 sigma sigma S0(mono) + 0.006, and sigma S0(1, 2, 3) = 0.723 x sigma sigma S0(mono) - 0.006. These are revealed to be successful for the estimation of the descriptors sigma S degrees for optional di- and tri-substituted benzene derivatives, and meet the needs of the practical qualitative structure-activity relationships.

Determination of the novel quantitative structure-activity relationships descriptor .SIGMA.S.DEG. for di- and trisubstituted benzene derivatives.

For di-and trisubstituted benzene derivatives, the novel quantitative structure-activity relationships descriptors σS°, representing both dispersion and repulsion interaction energies, have been successfully determined from the linear relations of the observed absolute entropy S°298(g) for the above series against those of monosubstituted benzene derivatives, Unknown values of S°298(g) could be estimated statistically by inter- or extrapolation of the linear relations, and numerous kinds of descriptors, σS° for ortho-, meta-, para-disubstituted benzene derivatives, together with those of the trisubstituted series, having the same kind of substituent, have been presented.Their validities are supported by the results of evaluation of the gas liquid chromatography relative retention value logγ for disubstituted benzene derivatives.