Consensus Docking

A consensus molecular docking pipeline supporting multiple docking tools: Smina, Gnina, LeDock, and GOLD.

Features

• Multi-tool docking: Run one or more docking tools and compare results
• Adaptive exhaustiveness: For Smina, automatically find optimal exhaustiveness level
• RMSD analysis: Compare poses across different docking tools
• Automated workflow: Handles file conversions, pose extraction, and result parsing

Supported Docking Tools

Tool	Input Format
Smina	Receptor: PDBQT, Ligand: SDF
Gnina	Receptor: PDB, Ligand: SDF
LeDock	Receptor: PDB, Ligand: SDF
GOLD	Receptor: PDB, Ligand: SDF

Installation

1. Create Conda Environment

conda env create -f environment.yml
conda activate consensus_docking

Optional: If you need opencadd (for some pocket coordinate calculations):

pip install git+https://github.com/volkamerlab/opencadd.git

2. Install Docking Tools

You must install the docking tools separately:

• Smina: Download from https://sourceforge.net/projects/smina/
• Gnina: Download from https://github.com/gnina/gnina/releases
• LeDock: Download from http://www.lephar.com/software.htm
• GOLD: Requires license from CCDC

Usage

Basic Example

python consensus_docker.py \
  --receptor_pdb protein.pdb \
  --ligand_sdf ligand.sdf \
  --pocket_pdb pocket.pdb \
  --smina_path /path/to/smina \
  --gnina_path /path/to/gnina \
  --use_smina \
  --use_gnina \
  --outfolder results/

Key Arguments

• --receptor_pdb: Protein structure (PDB format)
• --receptor_pdbqt: Protein structure (PDBQT format, optional)
• --ligand_sdf: Ligand structure (SDF format)
• --pocket_pdb: Binding pocket definition (PDB format)
• --outfolder: Output directory

Tool Selection

• --use_smina: Enable Smina docking
• --use_gnina: Enable Gnina docking
• --use_ledock: Enable LeDock docking
• --use_gold: Enable GOLD docking

Smina-Specific Options

• --exhaustiveness 12: Search exhaustiveness (default: 12)
• --adaptive_exhaustiveness: Auto-find optimal exhaustiveness level
• --num_modes 20: Number of binding modes (default: 20)
• --num_threads 4: CPU threads for docking

Advanced Options

# Adaptive exhaustiveness for Smina
python consensus_docker.py \
  --receptor_pdb protein.pdb \
  --ligand_sdf ligand.sdf \
  --pocket_pdb pocket.pdb \
  --smina_path ./smina \
  --use_smina \
  --adaptive_exhaustiveness \
  --maximum_exhaustiveness 32 \
  --exhaustiveness_increment 8 \
  --convergence_rmsd_threshold 1.5 \
  --outfolder results/

# Skip RMSD calculation between tools
python consensus_docker.py \
  --receptor_pdb protein.pdb \
  --ligand_sdf ligand.sdf \
  --pocket_pdb pocket.pdb \
  --smina_path ./smina \
  --gnina_path ./gnina \
  --use_smina \
  --use_gnina \
  --skip_rmsd \
  --outfolder results/

Output Structure

results/
├── input/
│   ├── receptor.pdbqt
│   ├── receptor.mol2
│   └── ligand.mol2
├── smina/
│   ├── out.sdf              # All poses
│   ├── out.log              # Docking log
│   ├── out_1.sdf, ...       # Individual poses
│   ├── complex_1.pdb, ...   # Receptor-ligand complexes
│   └── results.csv          # Scores
├── gnina/
│   ├── out.sdf
│   ├── out.log
│   ├── out_1.sdf, ...
│   ├── complex_1.pdb, ...
│   └── results.csv          # Affinity, CNN pose score, CNN affinity
├── ledock/
│   └── ...
├── gold/
│   └── ...
└── log_*.log

Output Scores

Smina

• SMINA_Score: Binding affinity (kcal/mol)

Gnina

• GNINA_Affinity: Binding affinity (kcal/mol)
• GNINA_CNN_pose: CNN pose score
• GNINA_CNN_affinity: CNN affinity prediction

LeDock

• LeDock_Score: Binding score

GOLD

• Gold.PLP.Fitness: PLP fitness score

Notes

• Gnina does not support adaptive exhaustiveness
• LeDock and GOLD require receptor in PDB format
• Smina requires receptor in PDBQT format (auto-converted if PDB provided)
• RMSD calculation requires results from multiple tools