A novel contrastive learning framework that aligns molecular and protein representations for accelerated virtual drug screening, achieving state-of-the-art performance across 14 benchmark datasets.
DrugCLIP introduces several novel contributions to molecular representation learning and virtual screening.
Novel cross-modal contrastive objective that jointly trains SMILES and graph-based molecular encoders with protein pocket representations.
Standardized evaluation across 14 datasets spanning ADMET, binding affinity, toxicity, and activity prediction with unified metrics and protocols.
Hierarchical representation capturing atom-level, fragment-level, and global molecular semantics through a novel graph transformer architecture.
Pre-trained representations transfer effectively to unseen protein targets without fine-tuning, enabling rapid screening of novel therapeutic areas.
Direct prediction of binding affinities from contrastive embeddings, rivaling physics-based docking methods at a fraction of the computational cost.
Integrated uncertainty-aware active learning loop that iteratively selects the most informative molecules for wet-lab validation, reducing experimental cycles.
DrugCLIP's architecture combines advances in contrastive learning, graph neural networks, and protein language models.
DrugCLIP accelerates multiple stages of the drug discovery pipeline.
Rapidly screen ultra-large virtual libraries (billions of compounds) against novel protein targets to identify promising hit molecules for further optimization.
Predict absorption, distribution, metabolism, excretion, and toxicity properties early in the pipeline to prioritize drug-like candidates and reduce attrition.
Guide medicinal chemistry by predicting how structural modifications affect binding affinity and selectivity, accelerating the lead optimization cycle.
From open-source code to enterprise API access, choose the level of access that fits your research needs.
Access the DrugCLIP model, benchmark suite, and codebase to bring AI-powered virtual screening into your research pipeline.
Join leading organizations using drugclip-paper
Get Started Free