gdock 🦀

Information-driven protein-protein docking using a genetic algorithm

IMPORTANT

🚧 gdock is currently undergoing a full Rust-rewrite 🚧

In its current state it is not yet production ready! The code still needs to be reviewed and the protocol needs to be benchmarked.

The older v1.1.0 python version is still available in the releases.

gdock is a fast, efficient protein-protein docking tool written in Rust that uses restraints and energy components to guide the docking process. It combines a genetic algorithm with physics-based scoring to find optimal protein-protein complexes.

Features

• 🚀 Fast: Genetic algorithm with early stopping and elitism
• 🎯 Information-driven: Uses residue restraints to guide docking
• 📊 Flexible scoring: Configurable energy weights (VDW, electrostatics, desolvation, restraints)
• 📈 Quality metrics: Optional DockQ calculation when reference structure is provided
• 🦀 Rust-powered: Memory-safe and high-performance

Installation

git clone https://github.com/gdocking/gdock
cd gdock
cargo build --release

The binary will be available at ./target/release/gdock.

Usage

Basic Usage (Docking Mode)

Run the full genetic algorithm docking:

gdock \
  --receptor data/A.pdb \
  --ligand data/B.pdb \
  --restraints 933:6,936:8,940:42,941:44,946:45,950:46

This will:

• Dock the ligand onto the receptor using the genetic algorithm
• Use the specified residues to create restraints
• Optimize based on energy scoring
• Output best_by_score.pdb

With Reference Structure (DockQ Mode)

When you have a reference/native structure for validation:

This is only useful in parametrization scenarios or when you are comparing gdock. Since in a real scenario you will not have the actual reference.

gdock \
  --receptor data/A.pdb \
  --ligand data/B.pdb \
  --restraints 933:6,936:8,940:42,941:44,946:45,950:46 \
  --reference data/2oob.pdb

This will:

• Calculate DockQ, RMSD, iRMSD, and FNAT during optimization
• Output both best_by_score.pdb and best_by_dockq.pdb

Score-Only Mode

Calculate energy components without running the GA:

gdock \
  --receptor data/A.pdb \
  --ligand data/B.pdb \
  --restraints 933:6,936:8,940:42,941:44,946:45,950:46 \
  --score

This will quickly compute VDW, electrostatic, desolvation, and AIR energies for the given complex.

Custom Energy Weights

gdock \
  --receptor data/A.pdb \
  --ligand data/B.pdb \
  --restraints 933:6,936:8,940:42,941:44,946:45,950:46 \
  --w_vdw 1.0 \
  --w_elec 0.5 \
  --w_desolv 0.1 \
  --w_air 100.0

Input Format

PDB Files

• Receptor: PDB file containing the receptor protein (single chain)
• Ligand: PDB file containing the ligand protein (single chain)
• Reference (optional): PDB file containing the native complex (both chains)

Restraints

• Colon-separated list of pairs of residue numbers in receptor:ligand format: 933:6,936:8, etc
• Indicates which residues should be part of the interaction, based on experimental data or other information source
• Creates all pairwise restraints between receptor and ligand residues
• More restraints = stronger guidance but may be restrictive!

Output

• best_by_score.pdb: Complex with the best energy score
• best_by_dockq.pdb: Complex with the best DockQ (only with --reference)

Algorithm

gdock uses:

• Genetic Algorithm: Population of 150, elitism (top 5), tournament selection
• Energy Function: VDW + Electrostatics + Desolvation + AIR restraints
• Restraints: Flat-bottom potential (0-7Å) for specified residue pairs
• Early Stopping: Converges when no improvement for 10 generations

Citation

Coming soon.

License

See LICENSE file.