Apppublic

kowith/anyone-docking

sourceHugging Faceupdated 5mo agoView on Hugging Face

0likes

App README

<img src="https://raw.githubusercontent.com/nyelidl/anyone-docking/main/any-L.svg" width="32"> nyone can dock, everyone can do!

Anyone docking: Browser-based molecular docking — no installation required.

![Streamlit App](https://nyelidl.github.io/anyone-docking/)

Paste a SMILES, draw a structure, or upload a file. Pick a PDB or CIF. Dock in seconds.

Batch docking with 4 docking engines: ![Open In Colab](https://colab.research.google.com/drive/1e23e0145ja6zJi0HibA6_JBO7p78Xvyw?usp=sharing)

✨ What it does


🔬	Single & batch docking via AutoDock Vina 1.2.7
🏗️	Automatic receptor prep — download any PDB/CIF, strip solvent, add hydrogens
📄	PDB & mmCIF support — upload `.pdb` or `.cif` files, or download either format from RCSB (auto-fallback to CIF for large/newer entries)
🎯	Auto grid detection from co-crystal ligand centroid with XYZ axis & box overlay
✏️	3-way ligand input — SMILES text, file upload (.pdb), or draw in Ketcher
♻️	Redocking validation in both single & batch mode — dock the co-crystal ligand as a reference with RMSD vs crystal, score comparison, and reference line in plots
🧪	Bond-order correction — fixes PDBQT aromaticity artifacts before visualization
🗺️	Three 2D diagram engines in separate tabs — see below for full details
🖱️	Interactive drag mode — freely reposition residue labels in real time, export PNG (up to 600 dpi) or SVG
🔭	Binding pocket viewer — interacting residues (orange sticks) around the docked ligand, with toggleable residue labels and adjustable distance cutoff
🤖	AI-ready prompt — auto-filled context for GPT, Claude, Gemini, or DeepSeek; adapts based on whether redocking was performed
📊	3D viewers — animated multi-pose sweep, interactive pose selector, and dedicated binding pocket view
📁	One-click ZIP — all poses, bond-order-corrected SDFs, 2D diagrams, and score plot

🗺️ 2D Interaction Diagrams

Three tabs — each with a different rendering engine:

🧬 Anyone Can Dock 2D Diagram (default)

A custom PoseView-style SVG diagram rendered entirely locally (no server required).

Feature	Detail
8 interaction types	H-bond (distance shown on line), hydrophobic, π-π stacking, cation-π, ionic, metal coordination, halogen bond, H-bond to halogen
Geometry-based detection	All interactions detected from 3D coordinates — no server, works on Streamlit Cloud
ACS ChemDraw bond style	Bond widths, double-bond spacing, and wedge geometry follow ACS publication standards
Smart layout	Residue circles placed radially by natural interaction angle; simultaneous-delta push-apart prevents overlap
Interactive drag mode	Click 🖱 Interactive to reposition any residue circle — lines and distance labels update in real time
Export	↓ SVG (vector) · ↓ PNG at Screen (1×) / 150 dpi (2×) / 300 dpi (3×) / 600 dpi (4×)

🔬 RDKit 2D Diagram

Classic RDKit MolDraw2DSVG highlight-circle style.

Feature	Detail
Interaction types	H-bond / polar (blue) · Hydrophobic (green) · Other (pink)
Layout	RDKit's own force-field layout — residue pseudo-atoms added as `BondType.ZERO` bonds, no manual placement needed
Side-by-side	Docked pose (left) + co-crystal reference (right)
Export	PNG + SVG download under each diagram
AI prompt	Auto-filled prompt adapts to single diagram vs. comparison

⬇ PoseView

Download-only tab — no API calls made from this app.

File	Description
`receptor.pdb`	Cleaned receptor (hydrogens added)
`docked_pose.sdf`	Selected docked pose
`cocrystal.sdf`	Co-crystal ligand (converted from PDB if needed)

Upload these files manually at proteins.plus/poseview to generate a server-side PoseView diagram.

🤖 AI Prompt Section

Located below the 2D diagram. The prompt auto-adapts to the session state:

Scenario	Prompt content
Docked ligand only	Plain-language explanation of interactions + ready-to-use summary paragraph
With co-crystal reference (no redocking)	Comparison prompt · Reference: `ligand` co-crystallised in PDB `ID` (see 2D diagram)
With co-crystal reference (redocking performed)	Comparison prompt · Reference binding energy from re-docking included

Copy the prompt + a screenshot of your diagram into Claude, GPT-4o, or Gemini to get a plain-English explanation of your results. The prompt ends with a "Ready-to-use summary:" section — a 3–4 sentence paragraph ready to paste into a report or slide.

💻 Platform compatibility

Platform	Vina binary	OpenBabel	Status
Linux x86_64	✅ Auto-download	`apt install openbabel`	Fully supported (primary)
macOS Intel	✅ Auto-download	`brew install open-babel`	Fully supported
macOS Apple Silicon (M1–M4)	✅ Native arm64	`brew install open-babel`	Fully supported
Windows x86_64	✅ Auto-download	Installer	Supported (WSL2 recommended)
Streamlit Cloud	✅ Auto-download	via `packages.txt`	Fully supported
Google Colab	✅ Auto-download	`!apt install openbabel`	Fully supported

Easiest option: Use the hosted Streamlit app — no installation needed.

🏗️ Receptor input formats

Format	Source	Notes
PDB	Upload `.pdb` or download from RCSB	Standard format, works for most entries
mmCIF	Upload `.cif` / `.mmcif` or download from RCSB	Recommended for large structures or newer PDB entries that lack `.pdb` files

CIF files are automatically converted to PDB using a multi-strategy cascade: gemmi → OpenBabel → ProDy. If PDB download from RCSB fails, the app automatically falls back to CIF format.

🖥️ Ligand input modes

Mode	Description
SMILES string	Type or paste any valid SMILES
Upload file	`.pdb` — converted automatically
Draw in Ketcher	Full 2D chemical sketcher in the browser → SMILES exported automatically

♻️ Redocking validation

Available in both single and batch docking modes:

Feature	Description
Co-crystal reference docking	Dock the known co-crystal ligand alongside your candidate(s)
RMSD vs crystal	Heavy-atom RMSD calculated against the original crystal pose
Reference score line	Dashed line on the affinity plot for quick visual comparison
Pose confirmation	Browse reference poses, confirm which one to use as the baseline
Download	Export reference poses as SDF/PDBQT

🔭 3D visualization layers

Viewer	What you see
Receptor prep	Protein cartoon · co-crystal ligand (magenta) · docking grid box (cyan wireframe) · XYZ axis arrows
Animated pose viewer	All poses swept as frames · protein surface · co-crystal overlay
Interactive pose selector	Single selected pose · protein cartoon + surface · co-crystal overlay
Binding pocket view	Faint full-protein cartoon · docked pose (cyan) · interacting residues (orange sticks) · optional residue labels
Redocking browser	Reference ligand poses · crystal overlay · RMSD per pose

🖥️ Run locally

Linux (Ubuntu / Debian)

bash

sudo apt install python3.11 python3.11-venv openbabel libcairo2-dev libpangocairo-1.0-0 && \
git clone https://github.com/nyelidl/anyone-docking-local.git && \
cd anyone-docking-local && \
python3.11 -m venv venv && \
source venv/bin/activate && \
pip install -r requirements.txt && \
streamlit run app.py

macOS

bash

brew install python@3.11 open-babel cairo pango && \
git clone https://github.com/nyelidl/anyone-docking-local.git && \
cd anyone-docking-local && \
python3.11 -m venv venv && \
source venv/bin/activate && \
pip install -r requirements.txt && \
streamlit run app.py

Apple Silicon (M1/M2/M3/M4): Fully supported — the app auto-downloads the correct aarch64 Vina binary.

Windows

Recommended: Use WSL2 with Ubuntu and follow the Linux instructions above — it's the simplest and most reliable path.

For native Windows, install dependencies manually first:

1.OpenBabel — download the installer from openbabel.org and add it to PATH
2.Cairo & Pango — easiest via conda: conda install -c conda-forge cairo pango

Then:

bash

git clone https://github.com/nyelidl/anyone-docking-local.git && \
cd anyone-docking-local && \
python -m venv venv && \
venv\Scripts\activate && \
pip install -r requirements.txt && \
streamlit run app.py

All platforms

—Python 3.10+ required
—AutoDock Vina 1.2.7 binary is downloaded automatically on first launch (Linux, macOS Intel/ARM, Windows)
—The app auto-detects your OS and CPU architecture

Optional: CIF support

For best mmCIF → PDB conversion quality, install gemmi:

bash

pip install gemmi

Without gemmi, the app falls back to OpenBabel and ProDy (both already in the dependency stack).

Streamlit Cloud deployment

Place these files in your repo root:

anyone-docking/
├── app.py
├── core.py
├── requirements.txt   # Python packages
└── packages.txt       # System apt packages (openbabel, libcairo2-dev, …)

📄 Citation

If you use this tool in research, please cite the relevant software and resources used in this workflow:

AutoDock Vina 1.2.7 Eberhardt et al., Journal of Chemical Information and Modeling, 2021 DOI: https://doi.org/10.1021/acs.jcim.1c00203

Anyone Can Dock Hengphasatporn, K.; Bunchuay T.; Duan, L.; Shigeta, Y., Journal of Chemical Information and Modeling, 2026 https://github.com/nyelidl/anyone-docking/

RDKit Landrum, G. (2023). RDKit: Open-source cheminformatics. https://www.rdkit.org

ProDy Bakan et al., Bioinformatics, 2011 DOI: https://doi.org/10.1093/bioinformatics/btr168

stmol Nápoles-Duarte et al., Frontiers in Molecular Biosciences, 2022 DOI: https://doi.org/10.3389/fmolb.2022.990846

Dimorphite-DL Ropp et al., Journal of Cheminformatics, 2019 DOI: https://doi.org/10.1186/s13321-019-0336-9

pKaNET Cloud Hengphasatporn, K. et al., J. Chem. Inf. Model. 2026, 66 (4), 1955–1963 DOI: https://doi.org/10.1021/acs.jcim.5c02852

gemmi (optional, for CIF support) Wojdyr, M., Journal of Open Source Software, 2022 DOI: https://doi.org/10.21105/joss.04200

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.