Developmental neurotoxicity (DNT) QSAR combination prediction model establishment and structural characteristics...

Area of Science:

• Environmental Chemistry
• Toxicology
• Computational Chemistry

Background:

• Rising incidence of neurodevelopmental disorders necessitates screening environmental chemicals for developmental neurotoxicity (DNT).
• Understanding the mechanisms of DNT is crucial for risk assessment and mitigation.

Purpose of the Study:

• To conduct the first qualitative structure-activity relationship (QSAR) study for DNT based on mammalian data.
• To identify structural characteristics associated with DNT compounds.

Main Methods:

• Utilized five classification algorithms and two feature selection methods to build DNT prediction models.
• Analyzed modeling descriptors and identified structure alerts (SAs).
• Combined three different models to enhance predictive performance and application domain.

Main Results:

• The best single model showed good predictive ability but a limited application domain.
• Combining models improved both Matthew's Correlation Coefficient (MCC) and application domain (AD) values.
• Identified electronic properties, van der Waals volume, and S, Cl, or P-containing substructures as associated with DNT.

Conclusions:

• This study provides a foundational QSAR model for predicting DNT of environmental chemicals.
• The findings contribute to a better understanding of DNT mechanisms.
• The identified structural features can guide future DNT assessment and chemical design.