Alignment Independent 3D-QSAR Modeling of Fullerene (C60) Solubility in Different Organic Solvents

Abstract

In this work, we have used two Molecular Interaction Fields (MIFs) based descrtipotrs, VolSurf and GRIND, as alignment-independent three-dimensional quantitative structure activity relationships (3D-QSAR) approaches to predict C60 solubilities in a diverse set of 132 organic solvents. GRIND methodology with fractional factorial design, and PLS analysis was carried out yielded a highly descriptive and predictive model. Genetic algorithm and successive projection algorithm (SPA) applied to feature selection and extract more informative VolSurf descriptors. A support vector machine (SVM) was used for model construction, and SPA-SVM-based VolSurf descriptors showed excellent performance in predicting C60 solubility. Validation, reliability and robustness of obtained models were evaluated by the prediction of external test sets, leave-one-out and progressive scrambling approach. The results confirmed that hydrophobic interactions besides steric effects are main factors influencing solubility of C60 in different organic solvents.