Quantitative Structure Activity Relationship Studies On The Binding Affinity Of Ligands To Thyroid Receptor

Environmental endocrine disruptors (estrogen disruptors, androgen disruptors and thyroid hormone disruptors, etc.) have caused serious risk to human and ecological health. Thyroid hormone disruptors (TDCs) can disrupt normal functions of the thyroid system and cause hazards to the growth, metabolism and development of humans and animals. Thus, it is of vital importance to screening the TDCs from a huge and ever-increasing number of chemicals, for the ecological risk assessment and environmental managing of the organic chemicals. However, there are more than 600 thousands of chemicals registered in Chemical Abstracts Service, and the number is increasing annually. As the experimental methods are costly and cannot screen the chemicals before large-scale production and consumption, it is necessary to develop theoretical computational methods for predicting the thyroid disrupting activities.In this study, a quantitative structure-activity relationship (QSAR) for the interactions of compounds and the thyroid hormone receptor a andβwas developed using genetic algorithm as variable reduction method coupled with multiple linear regression(MLR), partial least square regression (PLS) and support vector machine regression(SVM). Molecular descriptors were computed by Dragon, Discovery Studio and MOPAC2009 software. The built model was fully assessed by various validation methods, including internal and external validation, Y-randomization test, QUIK rule and chemical applicability domain. The validation results indicate that the QSAR model proposed had goodness-of-fit, robustness and predictivity. The SVM method was found to be the best of the three, resulting in accurate prediction, and the PLS method is better than the MLR method. For the SVM models, the determination coefficients are 0.87 and 0.84, the leave one out cross validation coefficients are 0.79 and 0.79 and the external correlation coefficients are 0.79 and 0.89, for TRa and TRβrespectively. Thus, the built QSAR models can be used to fast and accurately predict the binding affinity of compounds in the defined applicability domain. At the same time, the model proposed could also identify and provide some insight into what structural features are related to the biological activity of these compounds. The affinity of the studied compounds to thyroid hormone receptorαandβis highly related to the most negative formal charge of oxygen atom at the hydroxyl group. And molecular volume, molecular mass and polarizability are also important influencing factors.