Using support vector regression coupled with the genetic algorithm for predicting acute toxicity to the fathead minnow

Abstract

The potential toxicity of chemicals may present adverse effects to the environment and human health. The quantitative structure-activity relationship (QSAR) provides a useful method for hazard assessment. In this study, we constructed a QSAR model based on a highly heterogeneous data set of 571 compounds from the US Environmental Protection Agency, for predicting acute toxicity to the fathead minnow (Pimephales promelas). An approach coupling support vector regression (SVR) with the genetic algorithm (GA) was developed to build the model. The generated QSAR model showed excellent data fitting and prediction abilities: the squared correlation coefficients (r(2)) for the training set and the test set were 0.826 and 0.802, respectively. Only eight critical descriptors, most of which are closely related to the toxicity mechanism, were chosen by GA-SVR, making the derived model readily interpretable. In summary, the successful case reported here highlights that our GA-SVR approach can be used as a general machine learning method for toxicity prediction.

Ecotoxicological Modeling, Ranking and Prioritization of Pharmaceuticals Using QSTR and i‐QSTTR Approaches: Application of 2D and Fragment Based Descriptors

“…The descriptor is the modified drug‐like score belonging to the class of Drug‐like indices reported by Zheng et al. . As per equation 3, an increase in the value of this index causes a decrease in toxicity of pharmaceuticals.…”

Classification of 5-HT1A receptor agonists and antagonists using GA-SVM method

“…To select the most important descriptors and optimize model parameters simultaneously, an in-house program was coded in our laboratory using genetic algorithm (GA) [55] . GA is a method that randomly initializes a population of solutions and then improves it through repetitive operations of mutation, crossover and selection.…”

A joint optimization QSAR model of fathead minnow acute toxicity based on a radial basis function neural network and its consensus modeling

“…3 The fathead minnow (Pimephales promelas) is important as a biological model in aquatic toxicology studies as it represents one top trophic level of the aquatic food chain. 4 In previous studies, [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] the acute toxicity to the fathead minnow has become an important indicator to evaluate the hazards and risks of compounds in the aquatic environment. The published QSAR models for the acute toxicity of fathead minnow can mainly be divided into three types: (1) local models based on modes of acute toxic action, [3][4][5][6][7] (2) global models based on non-congeneric compounds, [2][3][4][5][7][8][9][10][11][12][13][15][16][17][18][19] and (3) consensus models composed of multiple global or local models.…”

“…4 In previous studies, [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] the acute toxicity to the fathead minnow has become an important indicator to evaluate the hazards and risks of compounds in the aquatic environment. The published QSAR models for the acute toxicity of fathead minnow can mainly be divided into three types: (1) local models based on modes of acute toxic action, [3][4][5][6][7] (2) global models based on non-congeneric compounds, [2][3][4][5][7][8][9][10][11][12][13][15][16][17][18][19] and (3) consensus models composed of multiple global or local models. 4,7,14 The advantage of local models is that they usually have better statistical performance and offer a better interpretation of the mechanism.…”

“…So far, many global predictive models for fathead minnow have been published. [1][2][3][4][5][7][8][9][10][11][12][13][15][16][17][18][19] In these studies, a variety of algorithms, such as Random Forest (RF), 1 Multiple Linear Regression (MLR), 2,5 a Support Vector Machine (SVM), 9 and an Articial Neural Network (ANN) 15 were used to develop the models. Although some models achieved good statistical results in some aspects of model performance (R 2 ¼ 0.89-0.99 (R 2 is the square correlation coefficients of tting) 15 and the best value of Q ext 2 of models established in ref.…”