TOXRIC

TOXRIC provides a standardized, ML-ready repository of toxicological data to support compound toxicity assessment and predictive model development.

Key Features:

• Extensive Data Repository: Contains 113,372 compounds across 13 distinct toxicity categories and 1,474 toxicity endpoints covering both in vivo and in vitro data.
• Diverse Feature Types: Integrates 39 feature types including structural descriptors, target information, transcriptome profiles, and metabolic pathways to support multifactorial analyses.
• Machine Learning Integration: Supplies ML-ready datasets that can serve as inputs or outputs for predictive algorithms and supports selection of optimal feature types and molecular representations for endpoint prediction.
• Visualization Tools: Provides visualizations of molecular structures and benchmark data to aid interpretation of toxicological mechanisms and evaluation of baseline algorithms.

Scientific Applications:

• Early-stage toxicity screening: Enables identification of potential toxicity during compound and drug discovery using curated endpoint data and predictive features.
• Mechanistic toxicology: Supports analysis of toxicological mechanisms through integrated transcriptome, target, and pathway features.
• Computational method development: Serves as a benchmark resource for developing and refining ML-based toxicity prediction algorithms.
• Ecological impact assessment: Facilitates studies of environmental impacts of compounds using the repository's in vivo and in vitro toxicity endpoints.

Methodology:

Provides ML-ready datasets, benchmark datasets, and visualizations; supports selection of feature types and molecular representations for predictive modeling.