Dock a ligand into a protein online

What this involves

Molecular docking identifies where and how a small molecule might bind to a macromolecular target. It provides a structural hypothesis for ligand orientation and a weak ranking metric across compounds, rather than an accurate binding free energy. The agent manages receptor curation, ligand tautomer and conformer generation, site definition, and docking calculations directly on dedicated computing resources without requiring local software installation.

A dependable docking campaign depends on rigorous receptor preparation. Apo crystal structures or poorly modelled pockets often fail because side chains are packed for bulk solvent rather than ligand engagement. Structural waters bridging critical interactions, cofactors, and catalytic metals must be preserved to prevent docking into unphysical cavities, while the protonation and tautomeric states of both protein and small molecule must be assigned correctly at pH 7.4.

Scores from classical scoring functions correlate only weakly with experimental affinity and tend to favour larger molecules. Docking workflows executed by the agent enforce internal strain evaluations, ligand efficiency metrics, and obligatory positive redocking controls using crystal ligands to prevent false positives and unphysical binding poses.

How it goes

1. 1Receptor Selection and PreparationThe agent queries the PDB for holo structures solved at 2.5 Å or better, retaining essential cofactors, catalytic metals, and conserved bridging waters while pruning bulk solvent, protonating at pH 7.4, and setting histidine tautomers based on local hydrogen-bond networks.
2. 2Binding Site DefinitionThe search space is centred on the co-crystallised ligand coordinates or mapped via pocket detection tools like fpocket, defining a targeted bounding box typically 20 to 25 Å per side to avoid search dilution.
3. 3Ligand Conformer and Tautomer GenerationLigand protonation states are assigned at pH 7.4, unspecified stereocentres and tautomers are enumerated, and low-energy three-dimensional conformers are built using RDKit's ETKDGv3 and MMFF minimisation before conversion to PDBQT format with Meeko.
4. 4Docking Search and Pose VerificationSearches are executed with AutoDock Vina, SMINA, or generative models like DiffDock and Boltz-2 across multiple random seeds, followed by strain checks, contact evaluations, and ligand efficiency calculations.

Start it here

Ask in plain language. The agent works on a Linux machine of its own — it installs and runs what the job needs, keeps every file, and reads the results back with the numbers explained. Nothing to set up, and a long run continues after you close the page.

What to watch for

• Stripping catalytic metals or essential cofactors like haem, zinc, or SAM produces unphysical cavities and confident nonsense.
• Converting a docking score directly into Kd, Ki, or IC50 using RT ln K ignores scoring function errors of 2 to 3 kcal/mol.
• Calculating RMSD with superposition algorithms like GetBestRMS hides misdocked poses by reporting artificially low values for incorrect pockets.
• Discarding bridging structural waters deletes real pocket architecture, causing ligands to clash or fail to dock.
• Accepting top-scoring poses with internal strain above 25 kcal/mol selects scoring artefacts over realistic conformations.
• Treating single-seed blind docking as definitive creates false confidence, as large search boxes fail to converge without high exhaustiveness.

What it runs

Questions