Molecular docking online, free: dock a ligand against EGFR in 5 minutes (no install)

Molecular docking predicts how a small molecule binds a protein. The standard open-source engine, AutoDock Vina, is free — but getting it running locally means conda environments, receptor preparation, ligand prep with Meeko, and a fair amount of command-line. This guide skips all of that: we'll dock a known inhibitor against the EGFR kinase entirely in the browser, then screen a small library, in about five minutes.

What you need

• A target structure — we'll use EGFR (PDB 1M17).
• A ligand — we'll start with gefitinib, a known EGFR inhibitor.
• A free MolHub account (academic tier, no card).

Step 1 — Load the receptor

On the Receptors page, type a 4-character PDB ID (1M17) and click Import. MolHub fetches the structure and runs FPocket to detect druggable cavities, then suggests a docking box around the best pocket automatically — so you don't have to eyeball coordinates.

No crystal structure for your target? You can import an AlphaFold model by gene name instead.

Step 2 — Pick a ligand

Search the 2.9M-molecule library for your compound, paste a SMILES, or pull one from PubChem on demand. For this walkthrough, search gefitinib and open it. Every molecule is already RDKit-normalized and fingerprint-indexed, so it's immediately dockable.

Step 3 — Run the dock

From the molecule (or the Docking → New page), choose the EGFR receptor and submit. Under the hood MolHub does ligand 3D embedding, Meeko prep, and runs AutoDock Vina with the suggested box and exhaustiveness=8. A single dock finishes in well under a minute on CPU — no GPU needed.

Step 4 — Read the result honestly

You get a binding affinity in kcal/mol (more negative = stronger predicted binding), the ranked poses, and — importantly — the contact residues: the 3D viewer highlights which side chains the ligand touches. That's your trust signal. A score alone is easy to over-interpret; seeing the pose sit in the ATP pocket and hydrogen-bond to the hinge is what tells you the result is plausible.

Keep expectations calibrated: docking scores correlate only weakly with measured affinity. Treat docking as triage and hypothesis generation, not prediction. The right use is enrichment — ranking a library so your wet-lab effort goes to the most promising candidates first.

Step 5 — Screen a whole library (batch)

The real payoff is batch screening. Save a set of molecules as a dataset (search results, a PubChem similarity pull, or your own SMILES), then run 1 receptor × N molecules in a single batch. Results rank live by affinity; filter the top hits, export to CSV/SDF, and check ADMET / drug-likeness on each before committing. From there, one click finds purchasable analogs you can actually order.

Why browser-based

The tools (Vina, RDKit, Meeko, FPocket) are all free and open — the friction was always setup. Running them as a hosted service means no conda, no receptor-prep scripting, no GPU bill, and nothing to maintain. For a wet-lab scientist who just needs a ranked shortlist, that's the whole point.