Aug-MIA-QSPR modeling of the soil sorption of carboxylic acid herbicides.

Abstract

Soil sorption, described as logK OC (the logarithm of the soil/water partition coefficient normalized to organic carbon), was modeled using the augmented multivariate image analysis applied to quantitative structure-property relationship method for a series of 11 carboxylic acid herbicides. The statistical model was found to be highly predictive and reliable to estimate logK OC of other persistent organic pollutants in the soil, which are analogues of the carboxylic acids used in the QSPR model. The QSPR model derived from images corresponding to the chemical structures of the 11 herbicides is superior to the uniparameter model based on the octanol/water partition coefficient (logP) and, in addition, a pattern recognition model was built using principal component analysis. This model allowed clustering and separating compounds with low/moderate soil sorption from those with moderate/high soil sorption (compounds with the aryloxy function) using the second principal component.