Expert QSAR system for predicting the bioconcentration factor under the REACH regulation

Area of Science:

• Environmental Chemistry
• Computational Toxicology
• Quantitative Structure-Activity Relationship (QSAR)

Background:

• Expert systems integrate multiple models to leverage strengths and mitigate weaknesses.
• Existing Quantitative Structure-Activity Relationship (QSAR) classification schemes categorize compounds based on Bioconcentration Factor (BCF) prediction accuracy relative to the octanol-water partition coefficient (KOW).

Purpose of the Study:

• To develop an integrated expert system for predicting chemical Bioconcentration Factor (BCF).
• To combine the most effective BCF models for different compound classes identified by a QSAR scheme.
• To ensure model simplicity, transparency, and mechanistic interpretability for regulatory acceptance.

Main Methods:

• Utilized a previously established QSAR classification scheme to categorize compounds based on KOW-BCF relationships.
• Selected and integrated the best performing models (three VEGA models and one KOW-based equation) for each identified compound class.
• Evaluated the integrated model's performance and stability on unknown datasets.

Main Results:

• The rationalized integration of selected BCF models demonstrated stability and superior performance on unseen data compared to benchmark models.
• The developed expert system provides reliable BCF predictions across different compound categories.

Conclusions:

• The integrated expert system offers a robust and accurate method for predicting chemical Bioconcentration Factor (BCF).
• The emphasis on simplicity, transparency, and mechanistic interpretation facilitates its adoption within regulatory frameworks like REACH.