Comparative Study of QSAR/QSPR Correlations Using Support Vector Machines, Radial Basis Function Neural Networks, and Multiple Linear Regression

Abstract

Support vector machines (SVMs) were used to develop QSAR models that correlate molecular structures to their toxicity and bioactivities. The performance and predictive ability of SVM are investigated and compared with other methods such as multiple linear regression and radial basis function neural network methods. In the present study, two different data sets were evaluated. The first one involves an application of SVM to the development of a QSAR model for the prediction of toxicities of 153 phenols, and the second investigation deals with the QSAR model between the structures and the activities of a set of 85 cyclooxygenase 2 (COX-2) inhibitors. For each application, the molecular structures were described using either the physicochemical parameters or molecular descriptors. In both studied cases, the predictive ability of the SVM model is comparable or superior to those obtained by MLR and RBFNN. The results indicate that SVM can be used as an alternative powerful modeling tool for QSAR studies.

Linear Regression, Model Averaging, and Bayesian Techniques for Predicting Chemical Activities from Structure

“…Bayesian neural networks can provide extremely good predictive power under cross-validation scrutiny, but we prefer the interpretability afforded regression models which can lead to mechanistic understanding of how structure affects activity. A number of authors have tried to compare these different methods [67,68,69,70]. We also did not cover the vast field of binaryor categorical-valued activities [71,72,73], but even there the idea of Bayesian model averaging has improved predictive power [72].…”

Effect of 3D parameters on Antifungal Activities of Some Heterocyclic Compounds

“…In the present study we have considered 68 compounds having antifungal activities against Candida albicans which have been represented in terms of log 1/C values,where C is the molar concentration of a compound and log 1/C is the dependent variable that defines the biological parameter for QSAR equations.Activities of these compounds have been taken from literature [7]. The structural details of compounds used in the present study are given in Table 1.…”

A Machine Learning Approach to Designing Guidelines for Acute Aquatic Toxicity

“…However, Russom et al illustrated that chemical of the same class may act through different modes of action [9]. Furthermore, classical linear QSAR methods have been shown to have problems associated with overfitting and are not able to handle nonlinear relationships [10][11][12]. QSAR methods have evolved over time to include Modes of Action [13,14].…”