ToxiVerse: A Public Platform for Chemical Toxicity Data Sharing and Customizable Predictive Modeling

Area of Science:

• Computational toxicology and cheminformatics
• Drug development and environmental safety assessment
• Machine learning applications in chemical research

Background:

• Chemical toxicity assessment is crucial for drug development and environmental safety.
• Computational models offer an alternative to animal testing for chemical evaluation.
• There is a need for accessible machine learning tools in computational toxicology.

Purpose of the Study:

• To develop ToxiVerse, a public web-based platform for computational toxicology.
• To provide user-friendly tools for chemical bioprofiling and toxicity prediction.
• To support researchers lacking programming expertise in chemical safety evaluation.

Main Methods:

• Integrated platform with Bioprofiler, Database, and Cheminformatics modules.
• Bioprofiler module uses chemical-bioactivity data and machine learning for descriptors.
• Cheminformatics module enables data upload, curation, and Quantitative Structure-Activity Relationship (QSAR) model generation.

Main Results:

• ToxiVerse provides curated toxicity datasets for approximately 50,000 chemicals.
• The platform facilitates automatic chemical bioprofiling and QSAR model generation.
• Users can access data, perform chemical curation, and predict toxicity endpoints.

Conclusions:

• ToxiVerse empowers researchers with accessible tools for chemical toxicity assessment.
• The platform enhances chemical bioprofiling and machine learning-based toxicity prediction.
• ToxiVerse supports efficient chemical safety evaluation, reducing the need for animal testing.