QSAR Analysis of Platelet‐derived Growth Inhibitors Using GA‐ANN and Shuffling Crossvalidation

Investigation on the adsorption affinity of organic micropollutants on seaweed and its QSAR study

Seaweed, one of the most abundant biomaterials, can be used as a biosorbent to remove organic micropollutants. In order to effectively use seaweed to remove a variety of micropollutants, it is vital to rapidly estimate the adsorption affinity according to the types of micropollutants. Thus, the isothermal adsorption affinities of 31 organic micropollutants in neutral or ionic form on seaweed were measured, and a predictive model using quantitative structure-adsorption relationship (QSAR) modeling was developed. As a result, it was found that the types of micropollutants had a significant effect on the adsorption of seaweed, as expected, and QSAR modeling with a predictability (R²) of 0.854 and a standard error (SE) of 0.27 log units using a training set could be developed. The model's predictability was internally and externally validated using leave-one-out cross validation and a test set. Its predictability for the external validation set was R² = 0.864, SE = 0.171 log units. Using the developed model, we identified the most important driving forces of the adsorption at the molecular level: Coulomb interaction of the anion, molecular volume, and H-bond acceptor and donor, which significantly affect the basic momentum of molecules on the surface of seaweed. Moreover, in silico calculated descriptors were applied to the prediction, and the results revealed reasonable predictability (R² of 0.944 and SE of 0.17 log units). Our approach provides an understanding of the adsorption process of seaweed for organic micropollutants and an efficient prediction method to estimate the adsorption affinities of seaweed and micropollutants in neutral and ionic forms.

Comparative study on the extraction efficiency, characterization, and bioactivities of Bletilla striata polysaccharides using response surface methodology (RSM) and genetic algorithm-artificial neural network (GA-ANN)

This research established the optimal conditions for alkali-assisted extraction (AAE) of bioactive polysaccharides from Bletilla striata integrated with response surface methodology (RSM) and the genetic algorithm-artificial neural networks (GA-ANN). In comparison with RSM, the ANN model showed a relatively higher determination coefficient in the global output values (RSM: ANN = 0.9270: 0.9742) performing more satisfactorily in the validation. Under the optimum conditions (52 °C; 167 min, and 0.01 mol/L NaOH), the extraction yields, IC₅₀ of ABTS, and FRAP value were 29.53 ± 0.97 %, 3.41 mg/mL, and 39.11 μmol Fe²⁺/g, respectively. The results indicated that BSPs-A was mainly composed of glucose and mannose with small amounts of arabinose, galactose, and galacturonic acid, while possessed a molecular weight of about 305.94 kDa (Mw). The structural characterization of BSPs-A was initially characterized by FT-IR, SEM, and Congo red tests, which indicated that BSPs-A possessed a triple helix conformation of typical Bletilla striata polysaccharides. In addition, BSPs-A exhibited excellent antioxidant activity, which was further confirmed by a series of in vitro antioxidant activity assays including DPPH, ABTS, FRAP, and ORAC. After incubation in the BSA-glucose system for 15 days, BSPs-A showed inhibition of the advanced glycation end products (AGEs) formation for the first time.

QSAR Study of PARP Inhibitors by GA-MLR, GA-SVM and GA-ANN Approaches

Background:

The poly(ADP-ribose) polymerases (PARP) is a nuclear enzyme superfamily present in eukaryotes.

Methods:

In the present report, some efficient linear and non-linear methods including multiple linear regression (MLR), support vector machine (SVM) and artificial neural networks (ANN) were successfully used to develop and establish quantitative structure-activity relationship (QSAR) models capable of predicting pEC50 values of tetrahydropyridopyridazinone derivatives as effective PARP inhibitors. Principal component analysis (PCA) was used to a rational division of the whole data set and selection of the training and test sets. A genetic algorithm (GA) variable selection method was employed to select the optimal subset of descriptors that have the most significant contributions to the overall inhibitory activity from the large pool of calculated descriptors.

Results:

The accuracy and predictability of the proposed models were further confirmed using crossvalidation, validation through an external test set and Y-randomization (chance correlations) approaches. Moreover, an exhaustive statistical comparison was performed on the outputs of the proposed models. The results revealed that non-linear modeling approaches, including SVM and ANN could provide much more prediction capabilities.

Conclusion:

Among the constructed models and in terms of root mean square error of predictions (RMSEP), cross-validation coefficients (Q2 LOO and Q2 LGO), as well as R2 and F-statistical value for the training set, the predictive power of the GA-SVM approach was better. However, compared with MLR and SVM, the statistical parameters for the test set were more proper using the GA-ANN model.

Shuffling cross-validation-bee algorithm as a new descriptor selection method for retention studies of pesticides in biopartitioning micellar chromatography

Bee algorithm (BA) is an optimization algorithm inspired by the natural foraging behaviour of honey bees to find the optimal solution which can be proposed to feature selection. In this paper, shuffling cross-validation-BA (CV-BA) was applied to select the best descriptors that could describe the retention factor (log k) in the biopartitioning micellar chromatography (BMC) of 79 heterogeneous pesticides. Six descriptors were obtained using BA and then the selected descriptors were applied for model development using multiple linear regression (MLR). The descriptor selection was also performed using stepwise, genetic algorithm and simulated annealing methods and MLR was applied to model development and then the results were compared with those obtained from shuffling CV-BA. The results showed that shuffling CV-BA can be applied as a powerful descriptor selection method. Support vector machine (SVM) was also applied for model development using six selected descriptors by BA. The obtained statistical results using SVM were better than those obtained using MLR, as the root mean square error (RMSE) and correlation coefficient (R) for whole data set (training and test), using shuffling CV-BA-MLR, were obtained as 0.1863 and 0.9426, respectively, while these amounts for the shuffling CV-BA-SVM method were obtained as 0.0704 and 0.9922, respectively.

Prediction of the maximum nonseizure load of lubricant additives

Lubricating additives can improve the lubricant performance of base oil in reducing friction and wear and minimizing loss of energy. It is of great significance to study the relationship between chemical structures and lubrication properties of lubricant additives. This paper reports a quantitative structure–property relationship (QSPR) model of the maximum nonseizure loads ([Formula: see text]) of 79 lubricant additives by applying artificial neural network (ANN) based on the algorithm of backward propagation of errors. Six molecular descriptors appearing in the multiple linear regression (MLR) model were used as vectors to develop the ANN model. The optimal condition of ANN with network structure of [6-4-1] was obtained by adjusting various parameters by trial-and-error. The root-mean-square (rms) errors from ANN model are [Formula: see text] ([Formula: see text]) for the training set and [Formula: see text] ([Formula: see text]) for the test set, which are superior to the MLR results of [Formula: see text] ([Formula: see text]) for the training set and [Formula: see text] ([Formula: see text]) for the test set. Compared to the existing model for [Formula: see text], our model has better statistical quality. The results indicate that our ANN model can be applied to predict the [Formula: see text] values for lubricant additives.

In silico evaluation, molecular docking and QSAR analysis of quinazoline-based EGFR-T790M inhibitors

Mutated epidermal growth factor receptor (EGFR-T790M) inhibitors hold promise as new agents against cancer. Molecular docking and QSAR analysis were performed based on a series of fifty-three quinazoline derivatives to elucidate key structural and physicochemical properties affecting inhibitory activity. Molecular docking analysis identified the true conformations of ligands in the receptor's active pocket. The structural features of the ligands, expressed as molecular descriptors, were derived from the obtained docked conformations. Non-linear and spline QSAR models were developed through novel genetic algorithm and artificial neural network (GA-ANN) and multivariate adaptive regression spline techniques, respectively. The former technique was employed to consider non-linear relation between molecular descriptors and inhibitory activity of quinazoline derivatives. The later technique was also used to describe the non-linearity using basis functions and sub-region equations for each descriptor. Our QSAR model gave a high predictive performance [Formula: see text] and [Formula: see text]) using diverse validation techniques. Eight new compounds were designed using our QSAR model as potent EGFR-T790M inhibitors. Overall, the proposed in silico strategy based on docked derived descriptor and non-linear descriptor subset selection may help design novel quinazoline derivatives with improved EGFR-T790M inhibitory activity.

Bee algorithm and adaptive neuro-fuzzy inference system as tools for QSAR study toxicity of substituted benzenes to Tetrahymena pyriformis

A quantitative structure-activity relationship (QSAR) was developed to predict the toxicity of substituted benzenes to Tetrahymena pyriformis. A set of 1,497 zero- to three-dimensional descriptors were used for each molecule in the data set. A major problem of QSAR is the high dimensionality of the descriptor space; therefore, descriptor selection is one of the most important steps. In this paper, bee algorithm was used to select the best descriptors. Three descriptors were selected and used as inputs for adaptive neuro-fuzzy inference system (ANFIS). Then the model was corrected for unstable compounds (the compounds that can be ionized in the aqueous solutions or can easily metabolize under some conditions). Finally squared correlation coefficients were obtained as 0.8769, 0.8649 and 0.8301 for training, test and validation sets, respectively. The results showed bee-ANFIS can be used as a powerful model for prediction of toxicity of substituted benzenes to T. pyriformis.

The use of Bayesian nonlinear regression techniques for the modelling of the retention behaviour of volatile components of Artemisia species

The main aim of this work was to assess the ability of Bayesian multivariate adaptive regression splines (BMARS) and Bayesian radial basis function (BRBF) techniques for modelling the gas chromatographic retention indices of volatile components of Artemisia species. A diverse set of molecular descriptors was calculated and used as descriptor pool for modelling the retention indices. The ability of BMARS and BRBF techniques was explored for the selection of the most relevant descriptors and proper basis functions for modelling. The results revealed that BRBF technique is more reproducible than BMARS for modelling the retention indices and can be used as a method for variable selection and modelling in quantitative structure-property relationship (QSPR) studies. It is also concluded that the Markov chain Monte Carlo (MCMC) search engine, implemented in BRBF algorithm, is a suitable method for selecting the most important features from a vast number of them. The values of correlation between the calculated retention indices and the experimental ones for the training and prediction sets (0.935 and 0.902, respectively) revealed the prediction power of the BRBF model in estimating the retention index of volatile components of Artemisia species.

Quantitative structure-activity relationship analysis of human neutrophil elastase inhibitors using shuffling classification and regression trees and adaptive neuro-fuzzy inference systems

The purpose of this study was to develop quantitative structure-activity relationship models for N-benzoylindazole derivatives as inhibitors of human neutrophil elastase. These models were developed with the aid of classification and regression trees (CART) and an adaptive neuro-fuzzy inference system (ANFIS) combined with a shuffling cross-validation technique using interpretable descriptors. More than one hundred meaningful descriptors, representing various structural characteristics for all 51 N-benzoylindazole derivatives in the data set, were calculated and used as the original variables for shuffling CART modelling. Five descriptors of average Wiener index, Kier benzene-likeliness index, subpolarity parameter, average shape profile index of order 2 and folding degree index selected by the shuffling CART technique have been used as inputs of the ANFIS for prediction of inhibition behaviour of N-benzoylindazole derivatives. The results of the developed shuffling CART-ANFIS model compared to other techniques, such as genetic algorithm (GA)-partial least square (PLS)-ANFIS and stepwise multiple linear regression (MLR)-ANFIS, are promising and descriptive. The satisfactory results r2p = 0.845, Q2(LOO) = 0.861, r2(L25%O) = 0.829, RMSE(LOO)  = 0.305 and RMSE(L25%O)  = 0.336) demonstrate that shuffling CART-ANFIS models present the relationship between human neutrophil elastase inhibitor activity and molecular descriptors, and they yield predictions in excellent agreement with the experimental values.

QSAR modelling of integrin antagonists using enhanced Bayesian regularised genetic neural networks

Bayesian regularised genetic neural network (BRGNN) has been used for modelling the inhibition activity of 141 biphenylalanine derivatives as integrin antagonists. Three local pattern search (PS) methods, simulated annealing and threshold acceptance were combined with BRGNN in the form of a hybrid genetic algorithm (HGA). The results obtained revealed that PS is a suitable method for improving the ability of BRGNN to break out from the local minima. The proposed HGA technique is able to retrieve important variables from complex systems and nonlinear search spaces for optimisation. Two models with 8-3-1 artificial neural network (ANN) architectures were developed for describing α₄β₇ and α₄β₁ modulatory activities of integrin antagonists. Monte Carlo cross-validation was performed to validate the models and Q₂ values of 0.75 and 0.74 were obtained for α₄β₇ and α₄β₁ inhibitory activities, respectively. The scrambling technique was used for sensitivity analysis of descriptors appearing in ANN models. Frequencies of repetition and sensitivity analysis of molecular descriptors revealed that 3D-Morse descriptors are influential factors for describing α₄β₇ inhibitory activity, while in the case of α₄β₁ inhibitory activity, the Randic shape index, the lowest eigenvalue of the Burden matrix and the number of rotatable bonds are important parameters.