Study for the binding affinity of thyroid hormone receptors based on machine learning algorithm

ABSTRACT

Long-term exposure of exogenous compounds to thyroid hormone receptors (TRs) may lead to thyroid dysfunction. Quantitative structure-activity relationship (QSAR) is expected to predicting the binding affinity of compounds to TR. In this work, two comprehensive and large datasets for TRα and TRβ were collected and investigated. Five machine learning models were established to predict the pIC50 of compounds. Meanwhile, the reliability of the models was ensured by a variety of evaluation parameters. The results showed that the support vector regression model exhibited the best robustness and external prediction ability (r²_train = 0.77, r²_test = 0.78 for TRα, r²_train = 0.78, r²_test = 0.80 for TRβ). We have proposed an appropriate mechanism for explaining the TR binding affinity of a compound. The molecular volume, mass, and aromaticity affected the activity of TRα. Molecular weight, electrical properties and molecular hydrophilicity played a significant role in the binding affinity of compounds to TRβ. We also characterized the application domain of the model. Finally, the obtained models were utilized to predict the TR binding affinities of 109 compounds from the list of endocrine disruptors. Therefore, this model is expected to be an effective tool for alerting the effects of exogenous compounds on the thyroid system.

KEYWORDS:

• Machine learning
• QSAR
• thyroid hormone receptor α
• thyroid hormone receptor β
• regression model

Acknowledgements

No funding was received to assist with the preparation of this manuscript

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed at: https://doi.org/10.1080/1062936X.2022.2100823

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.