Derisking Future Agrochemicals before They Are Made: Large-Scale In Vitro Screening for In Silico Modeling of Thyroid...

Area of Science:

• Computational toxicology and cheminformatics
• Agrochemical safety and drug discovery

Background:

• Thyroid peroxidase (TPO) inhibition is a key event in thyroid hormone disruption and toxicity.
• Developing safer agrochemicals requires identifying compounds that do not inhibit TPO.
• Existing TPO inhibitor datasets are limited in size and structural diversity.

Purpose of the Study:

• To create a robust in silico model for predicting TPO inhibition.
• To identify novel TPO inhibitors from a large, diverse chemical library.
• To support the design of safer agrochemicals by predicting off-target effects.

Main Methods:

• Conducted a large-scale in vitro screening of over 100,000 agrochemical compounds.
• Applied various machine learning techniques to develop predictive models for TPO inhibition.
• Compared the performance of the developed model against existing datasets like ToxCast.

Main Results:

• Identified over 6,000 structurally novel TPO inhibitors from the screened library.
• Developed a predictive in silico model with improved generalization due to diverse training data.
• Provided a dataset of 34,524 compounds, including inhibitors and inactives, for future research.

Conclusions:

• The developed in silico TPO inhibition model is a valuable tool for agrochemical research.
• The model enables prediction of TPO inhibition for virtual compounds, aiding early-stage discovery.
• The study provides a significant expansion of known TPO inhibitors and a powerful predictive resource.