deepmodeling/Uni-Mol
A series of deep-learning pre-trained models for 3D molecular representation, quantum chemical property prediction, and protein-ligand docking.

Uni-Mol provides universal 3D molecular representation learning through pre-trained transformer-based models. The suite includes Uni-Mol for molecule property and binding pose prediction, Uni-Mol+ for quantum chemical modeling and conformation generation, Uni-Mol Tools for automated property prediction, and Uni-Mol Docking for protein-ligand complex structure prediction. The models are based on 3D spatial graph architectures and rank among top performers on benchmarks like OGB-LSC and OC20.