Advancing Structure-Free Docking: FlowDock is Now Available on Vecura

What is FlowDock?

FlowDock is a state-of-the-art geometric flow-matching generative model designed for protein-ligand docking and binding affinity prediction. By coupling an ESMFold prior with a continuous normalizing flow, the model denoises an initial receptor structure toward a ligand-bound conformation. It effectively bridges the gap between protein sequence and structural binding insights, allowing researchers to explore molecular interactions without the need for pre-existing co-crystal structures.

What can users do with FlowDock on Vecura?

With FlowDock integrated into Vecura, researchers can:

• Predict 3D binding poses: Generate high-quality 3D bound protein-ligand complex structures from just a receptor amino-acid sequence and a ligand SMILES string.
• Estimate binding affinity: Obtain pKd-like scalar estimates to help rank ligand candidates based on their predicted binding strength.
• Perform blind docking: Carry out docking studies on novel targets that lack prior structural data (apo or holo templates).
• Customize inference: Fine-tune sampling parameters, such as the number of ODE integration steps and stochasticity, to balance computational speed with prediction accuracy.

What the output means

The output includes 3D structural PDB files representing the modeled protein-ligand complex and a table of predicted binding affinity scores. These results provide structural hypotheses and relative rankings for potential binders.

This output should be used to support scientific decision making. It does not replace experimental validation.

Why this matters

In traditional drug discovery, blind docking—predicting the binding of a molecule to a receptor of unknown structure—has long been a major bottleneck. FlowDock significantly lowers this barrier by eliminating the dependency on known templates, enabling faster screening and structural insight for a much broader range of biological targets.

By providing an end-to-end pipeline that generates both geometry and affinity, FlowDock empowers researchers to prioritize compounds more effectively in the early stages of the drug discovery process.

• Developed by: Morehead & Cheng (BioinfoMachineLearning)
• Source: Official GitHub Repository
• Reference: Morehead, A., & Cheng, J. (2025). FlowDock: Geometric Flow-Matching for Protein-Ligand Docking. Bioinformatics (ISMB 2025).