Genetic training of network using chaos concept: application to QSAR studies of vibration modes of tetrahedral halides

Application of a genetic algorithm and an artificial neural network for global prediction of the toxicity of phenols to Tetrahymena pyriformis

ABSTRACT

Genetic algorithm (multiparameter linear regression; GA-MLR) and genetic algorithm-artificial neural network (GA-ANN) global models have been used for prediction of the toxicity of phenols to Tetrahymena pyriformis. The data set was divided into 150 molecules for training, 50 molecules for validation, and 50 molecules for prediction sets. A large number of descriptors were calculated and the genetic algorithm was used to select variables that resulted in the best-fit to models. The six molecular descriptors selected were used as inputs for the models. The MLR model was validated using leave-one-out, leave-group-out cross-validation and external test set. A three-layered feed forward ANN with back-propagation of error was generated using six molecular descriptors appearing in the MLR model. Comparison of the results obtained using the ANN model with those from the MLR revealed the superiority of the ANN model over the MLR. The root mean square error of the training, validation, and prediction sets for the ANN model were calculated to be 0.224, 0.202, and 0.224 and correlation coefficients (r2) of 0.926, 0.943, and 0.925 were obtained. The improvements are because of non-linear correlations of the toxicity of the compounds with the descriptors selected. The prediction ability of the GA-ANN global model is much better than that of previously proposed models.

GRAPHICAL ABSTRACT

Application of genetic algorithm-support vector machine (GA-SVM) for prediction of BK-channels activity

The support vector machine (SVM), which is a novel algorithm from the machine learning community, was used to develop quantitative structure-activity relationship (QSAR) for BK-channel activators. The data set was divided into 57 molecules of training and 14 molecules of test sets. A large number of descriptors were calculated and genetic algorithm (GA) was used to select variables that resulted in the best-fitted for models. A comparison between the obtained results using SVM with those of multi-parameter linear regression (MLR) revealed that SVM model was much better than MLR model. The improvements are due to the fact that the activity of the compounds demonstrates non-linear correlations with the selected descriptors. Also distances between Oxygen and Chlorine atoms, the mass, the van der Waals volume, the electronegativity, and the polarizability of the molecules are the main independent factors contributing to the BK-channels activity of the studied compounds.

Particle swarm algorithm trained neural network for QSAR studies of inhibitors of platelet-derived growth factor receptor phosphorylation

The multilayer feed-forward artificial neural network (ANN) has been widely used in QSAR studying. Back-propagation algorithm (BP) and the use of evolutionary search as an ANN training method has some limitations associated with overfitting, local optimum problems and slow convergence rate. In this paper, particle swarm optimization (PSO) as a relatively new optimization technique has been used in ANN training. Compared to ANN trained by BP algorithm and evolutionary search, ANN training by PSO algorithm (PSONN) show satisfactory performance, converges quickly towards the optimal position and can avoid overfitting in some extent. The PSONN has been testified by using in QSAR modeling for inhibitory activity of 4-[4-(N-substituted (thio) carbamoyl)-1-piperazinyl]-6,7-dimethoxyquinazoline derivatives.

Support vector machine based training of multilayer feedforward neural networks as optimized by particle swarm algorithm: Application in QSAR studies of bioactivity of organic compounds

Multilayer feedforward neural networks (MLFNNs) are important modeling techniques widely used in QSAR studies for their ability to represent nonlinear relationships between descriptors and activity. However, the problems of overfitting and premature convergence to local optima still pose great challenges in the practice of MLFNNs. To circumvent these problems, a support vector machine (SVM) based training algorithm for MLFNNs has been developed with the incorporation of particle swarm optimization (PSO). The introduction of the SVM based training mechanism imparts the developed algorithm with inherent capacity for combating the overfitting problem. Moreover, with the implementation of PSO for searching the optimal network weights, the SVM based learning algorithm shows relatively high efficiency in converging to the optima. The proposed algorithm has been evaluated using the Hansch data set. Application to QSAR studies of the activity of COX-2 inhibitors is also demonstrated. The results reveal that this technique provides superior performance to backpropagation (BP) and PSO training neural networks.

The lifetime of CFC substitutes studied by a network trained with chaotic mapping modified genetic algorithm and DFT calculations

The hydrohaloalkanes have attracted much attention as potential substitutes of chlorofluorocarbons (CFCs) that deplete the ozone layer and lead to great high global warming. Having a short atmospheric lifetime is very important for the potential substitutes that may also induce ozone depletion and yield high global warming gases to be put in use. Quantitative structure-activity relationship (QSAR) studies were presented for their lifetimes aided by the quantum chemistry parameters including net charges, Mulliken overlaps, E(HOMO) and E(LUMO) based on the density functional theory (DFT) at B3PW91 level, and the C-H bond dissociation energy based on AM1 calculations. Outstanding features of the logistic mapping, a simple chaotic system, especially the inherent ability to search the space of interest exhaustively have been utilized. The chaotic mapping aided genetic algorithm artificial neural network training scheme (CGANN) showed better performance than the conventional genetic algorithm ANN training when the structure of the data set was not favorable. The lifetimes of HFCs and HCs appeared to be greatly dependent on their energies of the highest occupied molecular orbitals. The perference of the RMSRE comparing to RMSE as objective function of ANN training was better for the samples of interest with relatively short lifetimes. C(2)H(6) and C(3)H(8) as potential green substitutes of CFCs present relatively short lifetimes.

Hybridized particle swarm algorithm for adaptive structure training of multilayer feed-forward neural network: QSAR studies of bioactivity of organic compounds

The multilayer feed-forward ANN is an important modeling technique used in QSAR studying. The training of ANN is usually carried out only to optimize the weights of the neural network and without paying attention to the network topology. Some other strategies used to train ANN are, first, to discover an optimum structure of the network, and then to find weights for an already defined structure. These methods tend to converge to local optima, and may also lead to overfitting. In this article, a hybridized particle swarm optimization (PSO) approach was applied to the neural network structure training (HPSONN). The continuous version of PSO was used for the weight training of ANN, and the modified discrete PSO was applied to find appropriate the network architecture. The network structure and connectivity are trained simultaneously. The two versions of PSO can jointly search the global optimal ANN architecture and weights. A new objective function is formulated to determine the appropriate network architecture and optimum value of the weights. The proposed HPSONN algorithm was used to predict carcinogenic potency of aromatic amines and biological activity of a series of distamycin and distamycin-like derivatives. The results were compared to those obtained by PSO and GA training in which the network architecture was kept fixed. The comparison demonstrated that the HPSONN is a useful tool for training ANN, which converges quickly towards the optimal position, and can avoid overfitting in some extent.

Application ofab initio theory to QSAR study of 1,4-dihydropyridine-based calcium channel blockers using GA-MLR and PC-GA-ANN procedures

The usefulness of the quantum chemical descriptors, calculated at the level of the RHF theory using 6-31G basis set for QSAR study of 1,4-dihydropyridine-based calcium channel antagonist was examined. A data set containing 45 dihydropyridine derivatives with known activity was used. Multiple linear regressions combined with genetic algorithm for variable selection and an artificial neural network model combined with principal component analysis for dimension reduction and genetic algorithm for factor selection (PC-GA-ANN) were employed. Some multiparametric MLR equations with good statistical quality were obtained for different classes of dihydropyridine derivatives. The resulting equations suggest that the electronic properties of the atoms belonging to the backbone of the molecules as well as the conformation of the molecules affect the binding of these molecules with their receptor. In the PC-GA-ANN, The principal components of the descriptors data matrix were used as the input of the neural network and then genetic algorithm was applied to select the most relevant set of principal components. Two ANN models with five selected principal components were obtained. These models, which have high statistical qualities, can predict the activity of the molecules with prediction errors lower than +/-5%.