Predicting infinite dilution activity coefficients of hydrocarbons in water using ant colony optimization

Quantitative structure–property relationship (QSPR) studies based on ant colony optimization (ACO) were carried out for the prediction of infinite dilution activity coefficients of hydrocarbons, γ∞, in water at 298.15 K. Semi-empirical quantum-chemical calculations at AM1 level were used to find the optimum 3D geometry of the studied hydrocarbons and different descriptors (1514 descriptors) were calculated by the HyperChem and Dragon softwares. A major problem of QSPR is the high dimensionality of the descriptor space; therefore, descriptor selection is the most important step. In this paper, an ant colony optimization (ACO) algorithm is proposed to select the best descriptors (five descriptors). The root-mean-square deviation using multiple linear regression method for calibration and prediction sets were 0.40 and 0.39, respectively. The resulting data indicate that the proposed method can be used to predict the infinite dilution activity coefficients of hydrocarbons in water, and also reveal that the ACO is a useful tool for descriptor selection with nice performance.