Natural-language molecular search is live

Drug discovery, in your browser.

A 2.9-million-compound bioactivity library, any protein from AlphaFold, and 37 billion buyable analogs — screen, dock, and triage in your browser. No cluster, no licenses, no installs. SMILES to docking score in under three minutes.

Free academic tier · No credit card

app.molhub.bio/docking/batch
Batch screen · EGFR (1M17)
1Gefitinibbest
−7.71 kcal/mol
2Erlotinib
−7.42 kcal/mol
3CHEMBL325894
−7.39 kcal/mol
4CHEMBL265562
−7.03 kcal/mol
3D pose · contact residues

Built on open science

ChEMBLPubChemAlphaFold DBEnamine REALZINC-22RCSB PDBRDKitAutoDock Vina

The whole campaign, one workspace.

Find candidates, screen them against your target, and validate the binding mode — without stitching together five tools.

Batch docking

Screen an entire library against one receptor in a single run. Ranked by Vina affinity, filterable, one click to CSV or a curated dataset.

Natural-language search

“Quinazoline EGFR inhibitors, MW < 450, good QED.” Claude turns plain English into a precise SMARTS + property query and runs it.

3D binding mode

Rotate the receptor, overlay every pose, and see exactly which residues the ligand contacts — your trust signal, not just a number.

Bring your own

Upload your SMILES library or a raw PDB and dock against your own structures. Your data stays in the EU.

Cross-database search

Tanimoto similarity, SMARTS substructure, and property filters across ChEMBL and PubChem — Morgan-FP indexed.

Pocket detection

FPocket finds druggable cavities on any structure and suggests the docking box automatically — CPU only, no GPU bill.

2.9M
molecules indexed
200M+
AlphaFold structures
37B+
buyable analogs
< 3 min
SMILES → score
New · the data moat

Not a toy dataset.
The whole chemical universe.

We loaded the full bioactivity record, wired in on-demand fetch from the giants, and connected billions of buyable compounds — so you never run out of molecules or targets.

2.9M

ChEMBL — fully loaded

The entire ChEMBL bioactivity library, RDKit-normalized and fingerprint-indexed in our DB. Real IC50/Ki data, not just structures.

119M

PubChem — on demand

Type a name or a SMILES and pull any PubChem compound into your library instantly — by exact match or 2D similarity.

200M+

AlphaFold — any target

No crystal structure? Import the AlphaFold prediction by UniProt or gene name and dock against proteins PDB never solved.

37B+

Buyable analogs

One click finds purchasable / make-on-demand analogs of any hit across Enamine REAL, ZINC, and in-stock catalogs.

ADMET

Drug-likeness, built in

Lipinski, Veber, lead-likeness, GI/BBB and PAINS/Brenk structural alerts on every molecule — triage before you spend a dollar.

$0

No data bills

Built entirely on open science — ChEMBL, PubChem, AlphaFold, RDKit, AutoDock Vina. No SciFinder licence, no per-seat tax.

From target to buyable hit, in five steps.

Each step is one screen. No scripting, no conda, no cluster queue.

  1. 1

    Bring your target

    Receptors

    Import a PDB crystal by ID, an AlphaFold model by UniProt/gene name, or upload your own structure. FPocket finds the druggable cavity and suggests the box.

  2. 2

    Gather your compounds

    Library

    Search 2.9M ChEMBL molecules, pull from PubChem on demand, or paste your own SMILES. Save the set as a dataset.

  3. 3

    Batch dock

    Docking

    Screen the whole dataset against the receptor with AutoDock Vina in one run — ranked live by binding affinity.

  4. 4

    Triage the hits

    Validate

    See the 3D pose and exact contact residues, and read ADMET / drug-likeness flags on each candidate before committing.

  5. 5

    Find what's buyable

    Source

    One click surfaces purchasable analogs from billions of catalog compounds, with a link straight to vendor pricing.

What sets us apart

Most AI drug tools claim accuracy.
We publish ours.

We dogfood MolHub on real targets and report the numbers — limitations included. In a field full of hype, an honest benchmark is the differentiator.

Honest benchmarks

Reproducible studies with real ROC-AUC and the caveats — the trust signal a skeptical scientist actually wants.

Campaign-complete

Target → 2.9M library → batch dock → ADMET → buyable analogs in one workspace. No stitching five tools together.

Free for academics

.edu / .ac verified, generous limits, no card. Built on open science — no SciFinder licence, no per-seat tax.

EU data residency

Your structures and libraries stay in the EU — a trust point for European labs and biotech.

Simple pricing

Free for academics. Affordable for small biotech. No GPU surcharges.

Academic
Free.edu / .ac verified
  • 25 docking jobs / mo
  • 12 AI Copilot campaigns / mo
  • 2.9M-molecule library
  • Search · 3D viewer · NL query
Start free
Most popular
Basic
$99per month
  • 30 docking jobs / mo
  • 20 AI Copilot campaigns / mo
  • Batch screening
  • Upload your own libraries & PDBs
  • CSV / SDF export
Start Basic
Pro
$499per month
  • 200 docking jobs / mo
  • 200 AI Copilot campaigns + generation
  • Priority queue
  • Team workspaces
  • API access
Start Pro

Run your first docking job today.

Sign up free, paste a SMILES, pick a target, and get a real Vina score in minutes.