A similarity-based QSAR model for predicting acute toxicity towards the fathead minnow (Pimephales promelas)

Abstract

REACH regulation demands information about acute toxicity of chemicals towards fish and supports the use of QSAR models, provided compliance with OECD principles. Existing models present some drawbacks that may limit their regulatory application. In this study, a dataset of 908 chemicals was used to develop a QSAR model to predict the LC50 96 hours for the fathead minnow. Genetic algorithms combined with k nearest neighbour method were applied on the training set (726 chemicals) and resulted in a model based on six molecular descriptors. An automated assessment of the applicability domain (AD) was carried out by comparing the average distance of each molecule from the nearest neighbours with a fixed threshold. The model had good and balanced performance in internal and external validation (182 test molecules), at the expense of a percentage of molecules outside the AD. Principal Component Analysis showed apparent correlations between model descriptors and toxicity.

An in silico algal toxicity model with a wide applicability potential for industrial chemicals and pharmaceuticals

“…The positive coefficient of NdsCH in Equation 3 shows that toxicity increases with the electrophilic character. The present results are in concordance with a study where NdsCH was reported to explain toxicity toward fathead minnow [38]. The atom-pair descriptor CATS 2D is based on a 2-dimensional structure encoding topological information, where CATS refers to a chemically advanced template search and AP denotes a hydrogen bond acceptor, either positively charged or ionizable [39].…”

MaNGA: a novel multi-niche multi-objective genetic algorithm for QSAR modelling

“…It contains six MDs, and depending on the different thresholds that were set to compute the AD, it reached a R (Table 6), reached better performances on all the metrics, even thought, the models obtained by using the kNN and SVR regression models use a higher number of MDs, while the model obtained by using the linear regression method uses only five MDs. Furthermore, the applicability domain values or our models (95%-97%) are higher than those reported for the previous model (60%-80%) (Cassotti et al, 2015).…”

“…te . Only the model described in (Cassotti et al, 2015) was identified by using the same MDs used in this study, and evaluated also in terms of Q 2 , RMSE and AD, thus we choose this model for comparison. The model was obtained with a combination of genetic algorithm and knn regression model.…”

The kernel-weighted local polynomial regression (KwLPR) approach: an efficient, novel tool for development of QSAR/QSAAR toxicity extrapolation models

“…The acute toxicity (LC 50 96 h) of 908 diverse organic chemicals towards the fathead minnow (Pimephales promelas) were considered for the last case study [23]. Cassotti et al [23] employed non-linear kNN approach to develop the predictive QSAR model and they obtained a model with k = 6 (R 2 = 0.62; Q 2 cv = 0.61; Q 2 Ext = 0.61). The studied toxicity range is: 0.053 to 9.612.…”

“…Therefore, we have taken five different datasets of varying sizes from big, medium to small, and used diverse chemical classes of compounds for modeling purposes. Four datasets [20][21][22] were previously utilized to develop the high-quality QSAAR model using a common linear regression technique like MLR and one dataset [23] was employed to develop non-linear model implicating kNN technique. In present study, all five datasets were used to develop models utilizing the KwLPR approach.…”