UQDD: Uncertainty Quantification for Drug Discovery using Hybrid Models

Welcome to the UQDD documentation. This site covers installation, data pipeline, model training, configuration, metrics, and API reference for the UQDD Python package.

Paper

Combining Bayesian and Evidential Uncertainty Quantification for Improved Bioactivity Modeling

Highlights

• Hybrid UQ models: EOE (Ensemble of Evidential) and EMC (Evidential MC Dropout)
• Baselines: Deep ensembles, MC-Dropout, and probabilistic neural networks
• Papyrus++ datasets for xC50 and Kx bioactivity endpoints
• Comprehensive evaluation: performance, calibration, probabilistic scoring, decision utility

See the tags page for topics across the documentation.

Use the top navigation to get started.