A comparison of model performance for six quantitative structure‐activity relationship packages that predict acute toxicity to fish

Abstract

Abstract-Some regulatory programs rely on quantitative structure-activity relationship (QSAR) models to predict toxic effects to biota. Many currently existing QSAR models can predict the effects of a wide range of substances to biota, particularly aquatic biota. The difficulty for regulatory programs is in choosing the appropriate QSAR model or models for application in their new and existing substances programs. We evaluated model performance of six QSAR modeling packages: Ecological Structure Activity Relationship (ECOSAR), TOPKAT, a Probabilistic Neural Network (PNN), a Computational Neural Network (CNN), the QSAR components of the Assessment Tools for the Evaluation of Risk (ASTER) system, and the Optimized Approach Based on Structural Indices Set (OASIS) system. Using a testing data set of 130 substances that had not been included in the training data sets of the QSAR models under consideration, we compared model predictions for 96-h median lethal concentrations (LC50s) to fathead minnows to the corresponding measured toxicity values available in the AQUIRE database. The testing data set was heavily weighted with neutral organics of low molecular weight and functionality. Many of the testing data set substances also had a nonpolar narcosis mode of action and/or were chlorinated. A variety of statistical measures (correlation coefficient, slope and intercept from a linear regression analysis, mean absolute and squared difference between log prediction and log measured toxicity, and the percentage of predictions within factors of 2, 5, 10, 100, and 1,000 of measured toxicity values) indicated that the PNN model had the best model performance for the full testing data set of 130 substances. The rank order of the remainder of the models depended on the statistical measure employed. TOPKAT also had excellent model performance for substances within its optimum prediction space. Only 37% of the substances in the testing data set, however, fell within this optimum prediction space.

Review of (Quantitative) Structure–Activity Relationships for Acute Aquatic Toxicity

“…Papers that use expert systems for estimation of aquatic toxicity are reviewed below. Moore et al [95] published a comparative analysis on model performance of six software packages that predict acute toxicity to fish [ECOlogical Structure -Activity Relationships (ECOSAR), TOPKAT, two neural networks, ASsessment Tools for the Evaluation of Risk (ASTER) and OASIS]. A probabilistic neural network was found to be the best predictor for external chemicals.…”

The human cytochrome P450 sub-family: Transcriptional regulation, inter-individual variation and interaction networks

“…Prediction of toxicity as a result of the bio-activation of chemicals that occurs during metabolism is relatively straightforward, and can be achieved to a good degree of accuracy using in silico programs such as DERECK, TOPCAT or HazardExpert [5][6][7]. However and area of increasing concern are toxic reactions caused due to interactions between coadministered chemicals.…”

Neural networks for effect prediction in environmental and health issues using large datasets

“…These models were sufficiently successful to be used for the quantitative estimation of FHM values for a large number of DSL substances. Furthermore, a very detailed study employing a variety of statistical measures by Moore et al [9], and using an external test set of 130 substances derived from other sources, showed that the PNN results were superior in almost all aspects when compared to the other models predictions, including those of ECOSAR [3], ASTER [10], and TOPKAT [11]. These results clearly demonstrate both the general ability of neural network methods to be capable of modeling such diverse data sets and the specific ability of the PNN in doing so.…”

“…It has not yet been part of any comparative study, such as the one by Moore et al [9], however a graph showing the estimated and measured values for the training data set is provided in the manual of that program and indicates a high degree of correlation.…”