TorchDrug
DeepGraphLearning (Mila / Université de Montréal)
PyTorch-based ML platform for drug discovery covering graph neural networks, geometric deep learning, knowledge graphs, generative models, and retrosynthesis. Provides unified API for property prediction, generation, and synthesis planning.
Best For
Research prototyping across multiple drug discovery ML tasks in a unified framework
License
Open Source (Apache 2.0)
Strengths
- +Unified API for diverse tasks
- +GPU-accelerated graph operations
- +Built-in datasets and benchmarks
- +Apache 2.0 license
Limitations
- −Research-oriented (not production-hardened)
- −Smaller community than PyG
- −Documentation gaps
R&D Pipeline Coverage
Related Tools
Chemprop v2
MIT (Barzilay, Coley et al.)
Open-source D-MPNN library for molecular property prediction. The architecture underlying ADMET-AI and ADMETlab 3.0. Used in Halicin antibiotic discovery.
DeepPurpose
Huang et al. (Harvard / MIT)
Deep learning toolkit for drug-target interaction (DTI) prediction, compound property prediction, protein-protein interaction prediction, and drug-drug interaction prediction. Supports 15+ encoding methods and 5+ model architectures.
More in Retrosynthesis
RetroTRAE
KNU LCBC (Kyungpook National University)
Single-step retrosynthesis prediction using fragment-based tokenization of atomic environments and a Transformer architecture. Mimics chemical reasoning by learning changes in atom environments between products and reactants.
AiZynthFinder
AstraZeneca Molecular AI
Multi-step retrosynthetic planning tool using Monte Carlo tree search guided by neural network policies. Recursively breaks down target molecules into purchasable precursors. Production-used at AstraZeneca.
Stay updated on TorchDrug
Weekly newsletter covering AI tool releases, benchmarks, and what practitioners actually use.