Ferramentas in silico aplicadas ao estudo de ecotoxicidade de micropoluentes orgânicos persistentes

Abstract

Organic micropollutants (MPO) include a wide range of widely used applied substances that occur in low orders on the order of nanograms per liter (ng.L-1) to micrograms per liter (μg.L-1). Although they occur at low casualties, MPO are continuously released and widely detected in the environment. The potential risk of the presence of many MPO has not yet been identified or regulated. To establish regulatory limits for MPO, additional research on the biological responses of these compounds is of particular importance. Given this scenario, the present work used computational predictive methods to provide theoretical support to analyze parameters of interest in the context of chemical safety assessment and to develop information necessary to fill data gaps to improve understanding of toxicity and potential danger of the MPO. In this study, computational approaches were developed to identify the molecular properties related to the acute toxicity mechanism of fish against compounds defined as MoA 1. To predict the toxicity of Daphnia magna and Pimephales promelas for compounds identified in different water sources in Brazil. To assess the chemical safety of leachate transformation products from landfill formed during the Fenton process. The models were developed using VolSurf + v descriptors. 1.0.7, Dragon v.7.04, RDKit, Morgan fingerprints (ECFP6-like), MACCS keys and Avalon fingerprints. Partial least squares (PLS, English, partial least squares) and Random Forest algorithm were used for data analysis. According to the best modeling practices, these models were developed and rigorously validated, demonstrating good predictive potential for all investigated parameters. The information reported here was used as a tool for screening and assessing the toxicity and chemical safety of organic micropollutants.