ProtoBind-Diff: A Structure-Free Diffusion Language Model for Protein Sequence-Conditioned Ligand Design
Preprint
Abstract
Designing small molecules that selectively bind to protein targets remains a central challenge in drug discovery. While recent generative models leverage 3D structural data to guide ligand generation, their applicability is limited by the sparsity and bias of structural resources. Here, we introduce ProtoBind-Diff, a structure-free masked diffusion model that conditions molecular generation directly on protein sequences via pre-trained language model embeddings. Trained on over one million active protein–ligand pairs from BindingDB, ProtoBind-Diff generates chemically valid, novel, and target-specific ligands without requiring structural supervision. In extensive benchmarking against structure-based models, ProtoBind-Diff performs competitively in docking and Boltz-1 evaluations and generalizes well to challenging targets, including those with limited training data. Despite never observing 3D information during training, its attention maps align with predicted binding residues, suggesting the model learns spatially meaningful interaction priors from sequence alone. This sequence-conditioned generation framework may unlock ligand discovery across the full proteome, including orphan, flexible, or rapidly emerging targets for which structural data are unavailable or unreliable.Keywords
This publication has 30 references indexed in Scilit:
- How Good are Current Pocket-Based 3D Generative Models?: The Benchmark Set and Evaluation of Protein Pocket-Based 3D Molecular Generative ModelsJournal of Chemical Information and Modeling, 2024
- AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python BindingsJournal of Chemical Information and Modeling, 2021
- Three-Dimensional Convolutional Neural Networks and a Cross-Docked Data Set for Structure-Based Drug DesignJournal of Chemical Information and Modeling, 2020
- A Deep Learning Approach to Antibiotic DiscoveryCell, 2020
- Analyzing Learned Molecular Representations for Property PredictionJournal of Chemical Information and Modeling, 2019
- GraphVAE: Towards Generation of Small Graphs Using Variational Autoencoders2018
- Quantifying the chemical beauty of drugsNature Chemistry, 2012
- Escape from Flatland: Increasing Saturation as an Approach to Improving Clinical SuccessJournal of Medicinal Chemistry, 2009
- Do Structurally Similar Molecules Have Similar Biological Activity?Journal of Medicinal Chemistry, 2002
- Simulation Analysis of Experimental Design Strategies for Screening Random Compounds as Potential New Drugs and AgrochemicalsJournal of Chemical Information and Computer Sciences, 1995