Multivariate Image Analysis Applied To Quantitative Structure-activity Relationship Research Articles

MIA-QSAR and docking studies of recent HPPD inhibitors.

Mesotrione is a triketone widely used as an inhibitor of the hydroxyphenylpyruvate deoxygenase (HPPD) enzyme. However, new agrochemicals should be continuously developed to tackle the problem of herbicide resistance. Two sets of mesotrione analogs have been recently synthesized and demonstrated successful phytotoxicity against weeds. Herein, these compounds were joined as to form a single data set and the HPPD inhibition of this enlarged library of triketones was modeled using multivariate image analysis applied to quantitative structure-activity relationships (MIA-QSAR). Further, docking studies were carried out to validate the MIA-QSAR findings and aid the interpretation of ligand-enzyme interactions responsible for the bioactivities (pIC50 ). The MIA-QSAR models based on van der Waals radii (rvdW ), electronegativity (ε), and rvdW /ε ratio as molecular descriptors were similarly and satisfactorily predictive (r2 ≥ 0.80, q2 ≥ 0.68, and r2 pred ≥ 0.68). Subsequently, the PLS regression parameters were applied to predict the pIC50 values of newly proposed derivatives, yielding a few promising agrochemical candidates. In addition, the calculated log P for most of these derivatives was found to be higher than that of mesotrione and the library compounds, indicating that they should be less prone to leach out and contaminate groundwater. MIA descriptors corroborated by docking studies were capable of reliably modeling the herbicidal activities of 68 triketones. Due to the substituent effects at the triketone framework, particularly of a nitro group in R3 , promising analogs could be designed. The P9 proposal demonstrated higher calculated activity and log P than the commercial mesotrione.

Computer-Assisted Improvement of Sulfonylureas with Antifungal Properties and Limited Herbicidal Activity: Potential Application in Forage Conservation.

This work reports studies at the molecular level of a series of modified sulfonylureas to determine the chemophoric sites responsible for their antifungal and herbicidal activities. For forage conservation, high antifungal potency and low phytotoxicity are required. A molecular modeling study based on multivariate image analysis applied to quantitative structure-activity relationship (MIA-QSAR) was performed to model these properties, as well as to guide the design of new agrochemical candidates. As a result, the MIA-QSAR models were reliable, robust, and predictive; for antifungal activity, the averages of the main validation parameters were r2 = 0.936, q2 = 0.741, and r2pred = 0.720, and for herbicidal activity, the model was very predictive (r2pred = 0.981 and r2m = 0.944). From the interpretation of the MIA-plots, 46 novel sulfonylureas with likely improved performance were proposed, from which 9 presented promising calculated selectivity indexes. Docking studies were performed to validate the QSAR predictions and to understand the interaction mode of the proposed ligands with the acetohydroxyacid synthase enzyme.

QSAR-Guided Proposition of N-(4-methanesulfonyl)Benzoyl-N'-(Pyrimidin-2-yl)Thioureas as Effective and Safer Herbicides.

Chlorinated agrochemicals play a major role in toxicity due especially to the labile C - Cl bond and high lipophilicity of organochlorines. In turn, urea and thiourea herbicides are widely used for weed control. A series of substituted N-benzoyl-N'-pyrimidin-2-yl thioureas has been recently synthesized and tested against Brassica napus L., demonstrating promising herbicidal activities, particularly for chlorinated derivatives. We have therefore modeled these activities using multivariate image analysis applied to quantitative structure-activity relationships (MIA-QSAR) to find out a significant and reliable correlation between measured and predicted inhibition of B. napus L. root growth (%) and, ultimately, to propose effective, non-chlorinated and/or less lipophilic N-(4-methanesulfonyl)benzoyl-N'-(pyrimidin-2-yl)thiourea candidates. The model was found to be predictive, giving an average r2pred in the external validation of 0.833. The predicted data for the proposed herbicides, interpreted in terms of MIA-plots of the chemical moieties responsible for bioactivity and supported by docking studies towards the photosystem II enzyme, suggest that substituents at both R1 and R2 positions modulate the agrochemical (R1 = Cl increases and R2 = OR decreases bioactivity) and environmental friendship (particularly with R2 = OH) performances of this class of compounds.

Double focus in the modelling of anti-influenza properties of 2-iminobenzimidazolines: pharmacology and toxicology

ABSTRACT Influenza affects millions of people globally and the appearance of drug-resistant strains is an ongoing problem. Therefore, this work reports the development of quantitative structure-activity relationship (QSAR) models to predict some biological properties of new 2-iminobenzimidazoline candidates for the treatment of the flu. A series of 2-iminobenzimidazoline derivatives with experimentally available values for cytotoxicity (pCC50) and anti-influenza activity (pIC50) was used for multivariate image analysis applied to QSAR (MIA-QSAR). The models were vigorously validated according to the best practices in QSAR and the chemical features responsible for the response variables were analysed based on MIA-plots, which assess the PLS regression coefficients and variable importance in projection scores. MIA descriptors encoding atomic properties (van der Waals radius and electronegativity) were capable of properly modelling the pCC50 and pIC50 data. The internally and externally validated models were used to predict the selectivity indexes (SI = pCC50/pIC50) of unprecedented analogues, which were designed upon analysis of the MIA-plots that show the substituent groups most affecting the biological data and by the combination of substructures of selected molecules. At least three promising anti-influenza candidates could be proposed from the predictive MIA-QSAR models.

Different approaches to encode and model 3D information in a MIA-QSAR perspective

Tridimensional information is a fundamental aspect for modelling and explaining biological/physicochemical properties. In this sense, the goal of this study was to explore different approaches for encoding this type of information into MIA-QSAR (Multivariate Image Analysis applied to Quantitative Structure-Activity Relationships) descriptors and to effectively model these new features. Originally, MIA-QSAR is a technique based on the treatment of 2D images of molecules. The approaches explored in this work were: (I) the use of 2D image projections of computationally optimized molecular geometries as a source of 3D information for a powerful machine learning method (support vector machine applied to regression); (II) the use of slice images obtained from the optimized molecules placed inside a theoretical box as a source of 3D descriptors for a multi-way regression method (trilinear PLS); and (III) the use of images viewed from different faces of the above box as an alternative source of 3D MIA-QSAR descriptors. These strategies were applied in three different data sets comprising anti-HCV, anti-SARS-CoV, and anti-HIV compounds. Satisfactory parameters for both internal and external validation were achieved in all three models, and the statistical results of correlation were at least similar to those earlier reported for these series of compounds. Nevertheless, the risk of chance correlation could not be excluded as demonstrated by y-randomization tests. Whereas the traditional MIA-QSAR method, that uses perfectly congruent, non-optimized geometries of pharmacophoric substructures as images, is more efficient than 3D MIA-QSAR, the latter uses tridimensional digital objects as descriptors for the first time in QSAR for regression purposes.

Conformational fingerprints in the modelling performance of MIA-QSAR: a case for SARS-CoV protease inhibitors

ABSTRACT Multivariate image analysis applied to quantitative structure–activity relationships (MIA-QSAR) has proved to be a high-performance 2D tool for drug design purposes. Nonetheless, MIA-QSAR strategy does not efficiently incorporate conformational information. Therefore, understanding the implications of including this type of data into the MIA-QSAR model, in terms of predictability and interpretability, seems a crucial task. Conformational information was included considering the optimised geometries and the docked structures of a series of disulfide compounds potentially useful as SARS-CoV protease inhibitors. The traditional analysis (based on flat-shape molecules) proved itself as the most effective technique, which means that, despite the undeniable importance of conformation for biomolecular behaviour, this type of information did not bring relevant contributions for MIA-QSAR modelling. Consequently, promising drug candidates were proposed on the basis of MIA-plot analyses, which account for PLS regression coefficients and variable importance in projection scores of the MIA-QSAR model.

3D perspective into MIA-QSAR: A case for anti-HCV agents.

Quantitative structure-activity relationship (QSAR) is a molecular modeling technique widely used in the discovery of novel drugs. Currently, there are many approaches for performing such analysis, which are commonly classified from 1D to 6D. 2D and 3D techniques are among the most exploited ones. Multivariate image analysis applied to QSAR (MIA-QSAR) is an example of 2D methodology that has presented a satisfactory performance in the generation of effective prediction models for biological/physicochemical properties. However, once this is a 2D method, conformational information is not explicitly considered, despite the well-known role of such type of information in explaining the biochemical behavior. Thus, the importance of conformation is undeniable, but the requirement of this information for QSAR analysis still needs to be studied. Therefore, this work aimed to provide a method for encoding 3D information into MIA-QSAR descriptors and analyze the consequences of this inclusion on this methodology. The strategy consisted in fully optimizing the molecular geometries of anti-HCV compounds and three-dimensionally align them before performing the MIA-QSAR procedure. As a result, it was possible to verify that this type of information does not improve the MIA-QSAR modeling performance; instead, the traditional procedure consisting of maximally congruent substructures generated a more reliable prediction model.

Discrete Fourier Transform-Based Multivariate Image Analysis: Application to Modeling of Aromatase Inhibitory Activity.

We recently generalized the formerly alignment-dependent multivariate image analysis applied to quantitative structure-activity relationships (MIA-QSAR) method through the application of the discrete Fourier transform (DFT), allowing for its application to noncongruent and structurally diverse chemical compound data sets. Here we report the first practical application of this method in the screening of molecular entities of therapeutic interest, with human aromatase inhibitory activity as the case study. We developed an ensemble classification model based on the two-dimensional (2D) DFT MIA-QSAR descriptors, with which we screened the NCI Diversity Set V (1593 compounds) and obtained 34 chemical compounds with possible aromatase inhibitory activity. These compounds were docked into the aromatase active site, and the 10 most promising compounds were selected for in vitro experimental validation. Of these compounds, 7419 (nonsteroidal) and 89 201 (steroidal) demonstrated satisfactory antiproliferative and aromatase inhibitory activities. The obtained results suggest that the 2D-DFT MIA-QSAR method may be useful in ligand-based virtual screening of new molecular entities of therapeutic utility.

Ten Years of the MIA-QSAR Strategy

It has been a decade since the introduction of the MIA-QSAR (acronym of Multivariate Image Analysis applied to Quantitative Structure Activity Relationship) method. While many successes have been achieved over the years, the path for the progressive development of this method has been far from straight; several hurdles have been encountered and corresponding solutions devised, some immediately and others gradually. This report offers a comprehensive treatise of the most relevant stages in the historical development of the MIA-QSAR strategy. New challenges and future prospects are also discussed.

Is molecular alignment an indispensable requirement in the MIA-QSAR method?

For a decade, the multivariate image analysis applied to quantitative structure-activity relationship (MIA-QSAR) approach has been successfully used in the modeling of several chemical and biological properties of chemical compounds. However, the key pitfall of this method has been its exclusive applicability to congeneric datasets due to the prerequisite of aligning the chemical images with respect to the basic molecular scaffold. The present report aims to explore the use of the 2D-discrete Fourier transform (2D-DFT) as a means of opening way to the modeling, for the first time, of structurally diverse noncongruent chemical images. The usability of the 2D-DFT in QSAR modeling of noncongruent chemical compounds is assessed using a structurally diverse dataset of 100 compounds, with reported inhibitory activity against MCF-7 human breast cancer cell line. An analysis of the statistical parameters of the built regression models validates their robustness and high predictive power. Additionally, a comparison of the results obtained with the 2D-DFT MIA-QSAR approach with those of the DRAGON molecular descriptors is performed, revealing superior performance for the former. This result represents a milestone in the MIA-QSAR context, as it opens way for the possibility of screening for new molecular entities with the desired chemical or therapeutic utility.

2D-Discrete Fourier Transform: Generalization of the MIA-QSAR strategy in molecular modeling

Adequate alignment of chemical structure images with respect to the basic scaffold in a series of chemical compounds constitutes an indispensable requirement for constructing multivariate images (MVIs) and subsequent molecular modeling using the Multivariate Image Analysis applied to Quantitative Structure–Activity Relationship (MIA-QSAR) approach. However, up to the moment, this alignment procedure has been manually performed, based on subjective ocular precision. The 2D-Discrete Fourier Transform (2D-DFT) is introduced as a strategy for creating a common base to construct MVIs for chemical structures using their magnitude spectra. The utility of magnitude spectra in QSAR studies has been evaluated through models for the antimalarial, anticancer and trichomonicidal activity of a series of 2, 5-diaminobenzophenone, 4-phenylpyrrolocarbazole and benzimidazole derivatives, respectively, yielding satisfactory results comparable to superior to those reported in the literature. It is anticipated that this strategy should enable the application of the MIA-QSAR approach to structurally diverse datasets other than a series of congeneric datasets.

Exploring MIA-QSARs' for antimalarial quinolon-4(1H)-imines.

A series of quinolon-4(1H)-imines have been recently discovered as antimalarials, targeting both the exoerythrocytic and erythrocytic stages of the parasite's development stages, which correspond to the phase of clinical symptoms. Endowed with chemical and metabolic stability, the quinolon-4(1H)- imines are thus presented as promissory dual-stage antimalarials. Three versions of multivariate image analysis applied to quantitative structure-activity relationship (MIA-QSAR) methods, namely traditional MIA-QSAR, augmented MIA-QSAR (aug-MIA-QSAR) and color-encoded aug-MIA-QSAR (aug- MIA-QSARcolor), were applied to model the antimalarial activities in this series of compounds. The multiple linear regression models indicated that the aug-MIA-QSAR method is more predictive and reliable than the others (R(2) = 0.8079, R(2)cv = 0.6647 and R(2)pred = 0.9691) for this series of compounds. The selected aug- MIA-QSAR descriptors were used for pattern recognition using discriminant analysis by partial least squares (PLS-DA), in order to separate compounds with low, moderate and high bioactivities.

Introducing new dimensions in MIA-QSAR: A case for chemokine receptor inhibitors

Multivariate image analysis applied to quantitative structure–activity relationships (MIA-QSAR) is a very simple correlative method that uses pixels (binaries) of chemical structures built from 2D viewer programs as descriptors; structural changes correspond to different pixel coordinates, which explain the variance in the bioactivities block. The MIA-QSAR method has shown to be predictive and capable of encoding some chemical information, but introduction of more descriptive information, such as atom size and colors to differentiate atom types, would improve predictability and interpretability. The bioactivities of a series of chemokine receptor (CCR2) inhibitors have been modeled using both conventional and atom color/size-dependent MIA-QSAR (namely aug-MIA-QSAR); the latter showed to be better. Moreover, the results were comparable to those obtained by 3D methodologies, indicating that 2D shape and substituent size are more significant descriptors than the conformational profiles required by field fit techniques.

Chemoface: a novel free user-friendly interface for chemometrics

Um software para analise multivariada foi desenvolvido com o objetivo de oferecer uma ferramenta computacional livre com interface grafica amigavel para pesquisadores, professores e estudantes com interesse em quimiometria. O Chemoface possui modulos capazes de resolver problemas relacionados com planejamento experimental, reconhecimento de padroes, classificacao e calibracao multivariada. E possivel obter uma variedade de graficos e tabelas para explorar os resultados. Neste trabalho, as principais funcionalidades do Chemoface sao exploradas usando estudos de caso reportados na literatura, tais como otimizacao de adsorcao de corante indigo em quitosana usando planejamento fatorial completo, analise exploratoria de amostras de propolis caracterizadas por ESI-MS (espectrometria de massas com ionizacao electrospray) usando PCA (analise de componentes principais) e HCA (analise hierarquica de agrupamentos), modelagem MIA-QSAR (analise multivariada de imagem aplicada a relacoes quantitativas estrutura-atividade) para predicao de parâmetro cinetico relacionado a atividade de peptideos contra dengue usando PLS (metodo de quadrados minimos parciais), e classificacao de amostras de vinho de diferentes variedades usando PLS-DA (PLS para analise discriminante). Todos os exemplos sao ilustrados com graficos e tabelas obtidos no Chemoface. A software for multivariate analysis was developed in order to provide a free computational tool with user-friendly graphical interface for researchers, professors and students with interest in chemometrics. Chemoface comprises modules that can solve problems related to experimental design, pattern recognition, classification and multivariate calibration. It allows obtaining a variety of high quality graphics and tables to explore results. In this work, the main features of Chemoface are explored using case studies reported in the literature, such as optimization of adsorption of indigo dye on chitosan using full factorial design, exploratory analysis of propolis samples characterized by ESI-MS (electrospray ionization-mass spectrometry) using PCA (principal component analysis) and HCA (hierarchical cluster analysis), MIA-QSAR (multivariate image analysis applied to quantitative structure activity relationship) modeling for the prediction of kinetic parameter related to activities of peptides against dengue using PLS (partial least squares), and classification of wine samples from different varieties using PLS-DA (PLS discriminant analysis). All examples are illustrated with graphs and tables obtained by means of Chemoface.

Desenhos de estruturas químicas correlacionam-se com propriedades biológicas: MIA-QSAR

Descriptors in multivariate image analysis applied to quantitative structure-activity relationship (MIA-QSAR) are pixels of bidimensional images of chemical structures (drawings), which were used to model the trichomonicidal activities of a series of benzimidazole derivatives. The MIA-QSAR model showed good predictive ability, with r2, q2 and rval. ext.2 of 0.853, 0.519 and 0.778, respectively, which are comparable to the best values obtained by CoMFA e CoMSIA for the same series. A MIA-based analysis was also performed by using images of alphabetic letters with the corresponding numeric ordering as dependent variables, but no correlation was found, supporting that MIA-QSAR is not arbitrary.

Exploring MIA-QSARs for farnesyltransferase inhibitory effect of antimalarial compounds refined by docking simulations

Two series of farnesyltransferase (FTase) inhibitors were grouped and their antimalarial activi-ties modeled by means of multivariate image analysis applied to quantitative structure-activity relationship (MIA-QSAR). A reliable model was achieved, with r2 for calibration, external prediction and leave-one-out cross-validation of 0.96, 0.87 and 0.83, respectively. Therefore, biological activities of congeners can be estimated using the QSAR model. The bioactivities of new compounds based on the miscellany of substructures of the two classes of FTase inhibitors were predicted using the MIA-QSAR model and the most promising ones were submitted to ADME (absorption, distribution, metabolism and excretion) and docking evaluation. Despite the smaller interaction energy of the two most promising, predicted compounds in comparison to the two most active compounds of the data set, one of the proposed structures did not violate any Lipinski’s rule of five. Therefore, it is either a potential drug or may drive synthesis of similar, improved compounds.

2D Chemical Drawings Correlate to Bioactivities: MIA-QSAR Modelling of Antimalarial Activities of 2,5-Diaminobenzophenone Derivatives

Two-dimensional chemical structures of a series of 2,5-diaminobenzophenone derivatives, some farnesyltransferase inhibitors, have shown to correlate with the corresponding antimalarial activities. The descriptors in this QSAR analysis are pixels of the chemical structures (two dimensional images) transformed into binaries and, therefore, the data variance explaining the variance in the activities block corresponds to the coordinates of each pixel in each molecule. This method, named multivariate image analysis applied to quantitative structure-activity relationship (MIA-QSAR), was applied to model the antimalarial activities of the titled compounds and the results were compared to well known three-dimensional QSAR techniques for the same class of compounds. In addition to the simplicity and high predictive performance of the MIA-QSAR modelling, this 2D image-based method has the potential of working well when equally simple, classical analysis fails. Overall, the present QSAR analysis based on 2D chemical drawings (constrained structures) dispensed conformational screening and 3D alignment to provide a reliable QSAR model; the physicochemical description about e.g. steric effects and chiral centers is all contained in the way in which substituents in a congeneric series are drawn, and the method can serve as a tool to introduce those who are planning to deal with drug design.

MIA-QSAR, PCA-ranking and least-squares support-vector machines in the accurate prediction of the activities of phosphodiesterase type 5 (PDE-5) inhibitors

Phosphodiesterase type-5 (PDE-5) is a key enzyme involved in the erection process. PDE-5 inhibitors, such as Sildenafil (ViagraTM), Vardenafil (LevitraTM) and Tadalafil (CialisTM), are used for the treatment of erectile dysfunction. Computer-assisted modelling of biological activities of PDE-5 inhibitors may make quantitative structure–activity relationship (QSAR) models useful for the development of safer (low side effects) and more potent drugs. The multivariate image analysis applied to QSAR (MIA-QSAR) method, coupled to partial least-squares (PLS) regression, has provided highly predictive QSAR models. Nevertheless, regression methods which take into account nonlinearity, such as least-squares support-vector machines (LS-SVMs), are supposed to predict biological activities more accurately than the usual linear methods. Thus, together with prior variable selection using principal component analysis ranking, MIA-QSAR and LS-SVM regression were applied to model the bioactivities of a series of cyclic guanine derivatives (PDE-5 inhibitors), and the results were compared with those based on linear methodologies. MIA-QSAR/LS-SVM was found to improve greatly the prediction performance when compared with MIA-QSAR/PLS, MIA-QSAR/N-PLS, CoMFA/PLS and CoMSIA/PLS models.

MIA-QSAR modelling of activities of a series of AZT analogues: bi- and multilinear PLS regression

The activities of a series of azidothymidine derivatives, compounds with anti-HIV potency, were computationally modelled using multivariate image analysis applied to quantitative structure–activity relationships (MIA-QSAR). Two regression methods were tested in order to find the best correlation between actual and predicted activities: bilinear (traditional) partial least squares (PLS), applied to the unfolded dataset, and multilinear PLS (N-PLS), applied to the three-way array. The predictive abilities of the PLS- and N-PLS-based models were found to be nearly equivalent, and both the methods derived QSAR models that are statistically superior to conventional QSAR, in which physicochemical descriptors and multiple linear regression were applied.

Bi- and Multilinear PLS Coupled to MIA-QSAR in the Prediction of Antifungal Activities of Some Benzothiazole Derivatives

The activities of a series of benzothiazole derivatives, some Candida albicans N-myristoyltransferase (Nmt) inhibitors, were modeled through MIA-QSAR (multivariate image analysis applied to quantitative structure-activity relationship) by using two different regression methods: N-PLS, applied to the three-way array, and PLS, applied to the unfolded array. Both models demonstrated excellent predictive ability, with results comparable to those obtained through 3D approaches. In order to compare the results obtained through MIA descriptors with the predictions of a classical 2D QSAR, some representative physicochemical descriptors were calculated and regressed against the experimental pIC50 values through multiple linear regression, demonstrating that MIA-QSAR was superior for this series of compounds.