A neural networks study of quinone compounds with trypanocidal activity

Abstract

This work investigates neural network models for predicting the trypanocidal activity of 28 quinone compounds. Artificial neural networks (ANN), such as multilayer perceptrons (MLP) and Kohonen models, were employed with the aim of modeling the nonlinear relationship between quantum and molecular descriptors and trypanocidal activity. The calculated descriptors and the principal components were used as input to train neural network models to verify the behavior of the nets. The best model for both network models (MLP and Kohonen) was obtained with four descriptors as input. The descriptors were T5 (torsion angle), QTS1 (sum of absolute values of the atomic charges), VOLS2 (volume of the substituent at region B) and HOMO-1 (energy of the molecular orbital below HOMO). These descriptors provide information on the kind of interaction that occurs between the compounds and the biological receptor. Both neural network models used here can predict the trypanocidal activity of the quinone compounds with good agreement, with low errors in the testing set and a high correctness rate. Thanks to the nonlinear model obtained from the neural network models, we can conclude that electronic and structural properties are important factors in the interaction between quinone compounds that exhibit trypanocidal activity and their biological receptors. The final ANN models should be useful in the design of novel trypanocidal quinones having improved potency.

Natural and Synthetic Naphthoquinones Active Against Trypanosoma Cruzi: An Initial Step Towards New Drugs for Chagas Disease

“…Electronic and structural properties are important factors in determining the interaction between quinone compounds with trypanocidal activity and their biological receptors. ANN models could be useful in the design of novel trypanocidal quinones having improved potency [132].…”

A universal scale of aromaticity for π‐organic compounds

“…In chemistry, the use of neural networks has further expanded into the analysis of spectral data, drug design, prediction of chemical reactivity and physical properties as well as the development of quantitative structure-activity relationship. [31][32][33][34][35][36] Some advantages of neural networks, useful for the purpose of this work, are as follows: (i) they are adaptative, i.e., they can take data and learn from it; (ii) they are essentially nonlinear; (iii) they are capable of generalization, i.e., they can correctly process information that only broadly resembles the original data training; (iv) they are fault-tolerant being capable of properly handling noisy or incomplete data.…”

Machine Learning Techniques and Drug Design

“…Both obtained nonlinear models predicted the biological activity with good agreement to experimental data and the authors concluded that the employed descriptors were important to describe the main interactions between the substances and the biological target [25]. The first QSAR model for substances with affinity to vesicular monoamine transporter-2 (VMAT-2) was constructed by Zheng et al [71] that used the MLP methodology.…”

“…This technique was successfully applied to chemical analysis problems using atomic absorption spectrometry [60] and to obtain the near-infrared spectral calibration of complex beverage samples [61]. Besides these applications, the SOM potentialities have been shown in the modeling toxicity [62] and design of novel trypanocidal quinone compounds [25]. Other use of SOM involves the separation of compounds in training and test.…”

“…A certain number of computational techniques have been found useful for the establishment of these relationships such as multiple linear regression (MLR) [11][12][13][14] and partial least squares (PLS) [15][16][17][18][19][20]. Recently, there is also a growing interest in the application of artificial neural networks (ANNs) and support vector machines (SVM) in the field of QSAR, as well as other molecular modeling approaches have been recognized as important tools in drug discovery [21][22][23][24][25][26][27][28][29]. The most common machine learning techniques (MLT) can be classified as supervised (such as Multilayer Perceptrons, Bayesian Neural Network, and Support Vector Machines), unsupervised (for example, Self-Organizing Maps) and hybrid models, such as Counter Propagation Neural Network (CPN), which comprises the advantages of supervised and unsupervised learning techniques.…”