Preprint
AI Safety & Alignment

ProTox 3.0: a webserver for the prediction of toxicity of chemicals

Priyanka Banerjee(Charité - Universitätsmedizin Berlin), Emanuel Kemmler(Charité - Universitätsmedizin Berlin), Mathias Dunkel(Charité - Universitätsmedizin Berlin), Robert Preißner(Charité - Universitätsmedizin Berlin)
April 22, 2024Nucleic Acids Research1,344 citations
PDFDOI

1.3k

Citations

192

Influential Citations

Nucleic Acids Research

Venue

2024

Year

Abstract

Interaction with chemicals, present in drugs, food, environments, and consumer goods, is an integral part of our everyday life. However, depending on the amount and duration, such interactions can also result in adverse effects. With the increase in computational methods, the in silico methods can offer significant benefits to both regulatory needs and requirements for risk assessments and the pharmaceutical industry to assess the safety profile of a chemical. Here, we present ProTox 3.0, which incorporates molecular similarity and machine-learning models for the prediction of 61 toxicity endpoints such as acute toxicity, organ toxicity, clinical toxicity, molecular-initiating events (MOE), adverse outcomes (Tox21) pathways, several other toxicological endpoints and toxicity off-targets. All the ProTox 3.0 models are validated on independent external sets and have shown strong performance. ProTox envisages itself as a complete, freely available computational platform for in silico toxicity prediction for toxicologists, regulatory agencies, computational chemists, and medicinal chemists. The ProTox 3.0 webserver is free and open to all users, and there is no login requirement and can be accessed via https://tox.charite.de. The web server takes a 2D chemical structure as input and reports the toxicological profile of the compound for each endpoint with a confidence score and overall toxicity radar plot and network plot.

Analysis

Why This Paper Matters

ProTox 3.0 addresses a critical need in chemical safety and drug development: rapid, accessible toxicity prediction. With over 1,300 citations, the ProTox series has become a standard tool for in silico toxicology. This version expands coverage to 61 endpoints, including modern Tox21 pathways and molecular-initiating events, making it relevant for regulatory risk assessment and pharmaceutical screening. The free, no-login web interface lowers barriers for researchers worldwide, democratizing access to computational toxicology.

Technical Contributions

  • Expanded endpoint coverage: 61 toxicity endpoints spanning acute toxicity, organ toxicity, clinical toxicity, molecular-initiating events, adverse outcome pathways (Tox21), and off-target effects.
  • Hybrid modeling: Combines molecular similarity (nearest-neighbor approaches) with machine-learning models (likely random forests or neural networks) to improve prediction robustness.
  • Validation rigor: All models validated on independent external sets, not just cross-validation, increasing confidence in generalizability.
  • User-friendly output: Provides confidence scores per endpoint, a radar plot for overall toxicity profile, and a network plot for visualizing relationships.

Results

The abstract states that all models "have shown strong performance" on independent external validation sets, but no specific metrics (e.g., AUC, accuracy, sensitivity) are provided. This lack of quantitative results limits direct comparison with other toxicity prediction tools. The paper likely contains detailed performance tables in the full text.

Significance

ProTox 3.0 represents a practical application of AI for safety assessment, directly impacting drug discovery pipelines and regulatory toxicology. By making advanced predictions freely available, it supports the 3Rs (Replacement, Reduction, Refinement) in animal testing. For the AI community, it demonstrates how combining classical cheminformatics (molecular similarity) with machine learning can yield production-ready tools. The platform's continued adoption (1344 citations) underscores its value as a benchmark in computational toxicology.

toxicity-predictionmachine-learningmolecular-similaritycomputational-toxicologywebserverchemical-safetydrug-discovery
Ai Safety

Related Papers