Overview

This project accompanies the paper "Combining Bayesian and Evidential Uncertainty Quantification (UQ) for Improved Bioactivity Modelling".

Goals:

• Provide reproducible pipelines for preparing Papyrus++ subsets and training hybrid UQ models
• Benchmark hybrid approaches (EOE, EMC) against baselines
• Offer metrics analysis including calibration, probabilistic scoring, and decision utility

Datasets & Endpoints:

• Papyrus++ curated subsets
• Endpoints: xC50 and Kx

Highlights:

• EOE (Ensemble of Evidential networks) and EMC (Evidential MC Dropout)
• Baselines: deep ensembles, MC-Dropout, probabilistic neural networks
• Tools: PyTorch 2.0+, RDKit, scikit-learn, Weights & Biases

See the User Guide for data, models, configuration, and metrics.