Machine learning-based prediction of toxicity of organic compounds towards fathead minnow

• DOI: 10.1039/d0ra05906d
• Publication Date: 2020-10-01T12:16:32Z
• Authors: Xingmei Chen; Limin Dang; Hai Yang; Xianwei Huang; Xinliang Yu
• Keywords: Pimephales promelas

Predicting the acute toxicity of a large dataset diverse chemicals against fathead minnows (Pimephales promelas) is challenging. In this paper, 963 organic compounds with towards were split into training set (482 compounds) and test (481 an approximate ratio 1 : 1. Only six molecular descriptors used to establish quantitative structure-activity/toxicity relationship (QSAR/QSTR) model for 96 hour pLC50 through support vector machine (SVM) along genetic algorithm. The optimal SVM (R 2 = 0.756) was verified using both internal (leave-one-out cross-validation) external validations. validation results (q int 0.699 q ext 0.744) satisfactory in predicting compared other models reported literature, although our has only data consisting 481 compounds.