QSAR: An In Silico Approach for Predicting the Partitioning of Pesticides into Breast Milk

Author(s): Suezana Agatonovic-Kustrin, David W. Morton, D. Celebic

Journal Name: Combinatorial Chemistry & High Throughput Screening
Accelerated Technologies for Biotechnology, Bioassays, Medicinal Chemistry and Natural Products Research

Volume 16 , Issue 3 , 2013

Become EABM
Become Reviewer


The aim of this study was to develop an in silico Quantitative Structure Activity Relationship (QSAR) model capable of predicting partitioning of pesticides into breast milk from their respective chemical structures. A large data set of 190 diverse compounds, including drugs and their active metabolites (87%), and pesticides (13%) with experimentally derived milk/plasma (M/P) ratios taken from the literature, was used to train, test and validate a predictive model. Each compound was encoded with 65 calculated chemical structure descriptors. Sensitivity analysis was then used to select a subset of the descriptors that best describe the transfer of pesticides into breast milk and Artificial neural networks modeling was applied to correlate selected descriptors (inputs) with the M/P ratio (output) in order to develop a predictive QSAR. The developed QSAR model included 26 molecular descriptors related to the molecular size, polarity and hydrogen binding capacity. Together with aromatic rings, these descriptors account for molecule’s size and hydrophobic interaction capabilities. The average correlation for the final model (incorporating training, testing, and validation) was 0.85. The developed model provides a useful method for predicting the M/P ratios of pesticides from just a sketch of their respective molecular structures. However, these predictions should only be used to assist in the evaluation of risk in conjunction with an assessment of the infant's response to a given drug/pesticide.

Keywords: ANNs, in silico modeling, milk-plasma partitioning, pesticides, QSAR

Rights & PermissionsPrintExport Cite as

Article Details

Year: 2013
Page: [223 - 232]
Pages: 10
DOI: 10.2174/1386207311316030007
Price: $65

Article Metrics

PDF: 22