Quantitative structure – activity relationships (QSARs) which are obtained with a representation of the molecular architecture via simplified molecular input-line entry system (SMILES) are applied to build up predictive models of acute toxicity of pesticides towards Daphnia magna. The acute toxicity towards Daphnia magna is an adequate measure of the ecological impact of various substances. The Monte Carlo technique is the basis to build up the above QSAR models. The statistical quality of suggested models is good: the best model is characterized by n = 103, R² = 0.76, RMSE = 0.91 (training set); n = 53, R² = 0.82, RMSE = 0.87 (validation set). The approach provides the mechanistic interpretation (e.g. aromaticity and branching of carbon skeleton are promoters of increase for toxicity towards Daphnia magna in the case of the examined set of pesticides). The approach is attractive to build up predictive models since instead of a large number of different molecular descriptors the corresponding model is based on solely one optimal descriptor calculated with SMILES and all necessary calculations can be done using the CORAL software available on the Internet (http://ww.insilico.eu/coral).

通过简化的分子输入线进入系统（SMILES）以分子结构的表示形式获得的定量结构-活性关系（QSAR）用于建立农药对水蚤的急性毒性的预测模型。对大型蚤（Daphnia magna）的急性毒性足以衡量各种物质的生态影响。蒙特卡洛技术是建立上述QSAR模型的基础。建议模型的统计质量良好：最佳模型的特征在于n = 103，R ²  = 0.76，RMSE = 0.91（训练集）；n = 53，R ² = 0.82，RMSE = 0.87（验证集）。该方法提供了机械学解释（例如，在一组农药中，碳骨架的芳香性和分支是增加对大型蚤的毒性的促进剂）。该方法对于建立预测模型很有吸引力，因为相应的模型不是基于大量不同的分子描述符，而是仅基于一个使用SMILES计算的最佳描述符，并且所有必要的计算都可以使用Internet上可用的CORAL软件完成（http： //ww.insilico.eu/coral）。