MagnusAGray/Tox21_MolBERT

Tox21 MolBERT Random Forest Classifier

This repository hosts a Hugging Face Space that provides a model submitted to the Tox21 Leaderboard.

Here, a Random Forest classifier is trained on MolBERT language embeddings computed from SMILES for the Tox21 dataset, and the trained model is provided for inference.

• —Input: a SMILES string of a small molecule
• —Output: 12 numeric values (probabilities between 0 and 1) for each of the toxicity endpoints of the Tox21 dataset.

This model was utilized in the paper Comparative Study of Molecular Descriptors and AI-Based Embeddings for Toxicity Prediction.