Deep learning driven QSAR model for environmental toxicology: Effects of endocrine disrupting chemicals on human health. (October 2019)

• Record Type: Journal Article
• Authors: Heo, SungKu; Safder, Usman; Yoo, ChangKyoo
• Is Part Of: Environmental pollution. Volume 253(2019)
• Page Start: 29
• Page End: 38
• Publication Date: 2019-10
• Subjects: Deep learning (DL) -- Emerging contaminants (ECs) -- Endocrine disrupting chemicals (EDCs) -- Sex hormone binding globulin (SHBG) -- Estrogen receptor (ER) -- Quantitative structure-activity relationship (QSAR); Pollution -- Periodicals; Pollution -- Environmental aspects -- Periodicals; Environmental Pollution -- Periodicals; Pollution -- Périodiques; Pollution -- Aspect de l'environnement -- Périodiques; Pollution -- Effets physiologiques -- Périodiques; Pollution; Pollution -- Environmental aspects; Periodicals; Electronic journals; 363.73
• DOI: 10.1016/j.envpol.2019.06.081
• Languages: English
• ISSNs: 0269-7491

Abstract

Abstract: Over 80, 000 endocrine-disrupting chemicals (EDCs) are considered emerging contaminants (ECs), which are of great concern due to their effects on human health. Quantitative structure-activity relationship (QSAR) models are a promising alternative to in vitro methods to predict the toxicological effects of chemicals on human health. In this study, we assessed a deep-learning based QSAR (DL-QSAR) model to predict the qualitative and the quantitative effects of EDCs on the human endocrine system, and especially sex-hormone binding globulin (SHBG) and estrogen receptor (ER). Statistical analyses of the qualitative responses indicated that the accuracies of all three DL-QSAR methods were above 90%, and greater than the other statistical and machine learning models, indicating excellent classification performance. The quantitative analyses, as assessed using deep-neural-network-based QSAR (DNN-QSAR), resulted in a coefficient of determination (R 2 ) of 0.80 and predictive square correlation coefficient (Q 2 ) of 0.86, which implied satisfactory goodness of fit and predictive ability. Thus, DNN was able to transform sparse molecular descriptors into higher dimensional spaces, and was superior for assessment qualitative responses. Moreover, DNN-QSAR demonstrated excellent performance in the discipline of computational chemistry by handling multicollinearity and overfitting problems. Graphical abstract: Image 1 Highlights: VIP and LASSO regression were implemented to select key molecular descriptors. DL-QSAR model was used to predict the responses of EDCs to SHBG and ER. DNN-QSAR model obtained Q 2 of 0.86 prediction performance. DNN-QSAR model showed accuracy of 97.0% for classification performance. Model performance of DL-QSAR models outperformed other conventional ML techniques. Abstract : DL-QSAR models were implemented to predict and classify the responses of the ER and SHBG to EDCs. The performances of proposed models were superior to conventional ML techniques.