<<<<<<< HEAD

comp_soft_architecture

MSC Life science 1. Semester computer and software architecture

Molecular Docking Workflow with Snakemake

A complete workflow for protein-ligand docking using Snakemake, based on the TeachOpenCADD tutorial T015 (https://projects.volkamerlab.org/teachopencadd/talktorials/T015_protein_ligand_docking.html).

Overview

This pipeline automates the molecular docking process:

1. Prepare structures - Extract protein and ligand from PDB
2. Convert to PDBQT - Format conversion for docking
3. Molecular docking - Run smina (AutoDock Vina fork) with 5 poses per ligand
4. Post-processing - Extract docking scores
5. Split poses - Separate 5 individual docking poses
6. Visualization - Generate 3D protein-ligand complex images using PyMOL for each pose found in the PDB file data

Requirements

• Python 3.10+
• Conda (Miniforge or Anaconda)

Setup

1. Clone the repository

git clone <your-repo-url>
cd dock_project

2. Create the conda environment

conda env create -f environment.yml
conda activate dock_env

This installs all dependencies:

• snakemake (workflow engine)
• openbabel (structure format conversion)
• pymol-open-source (3D visualization)
• rdkit (molecular toolkit)
• smina (docking engine)
• And supporting libraries

3. Add your input PDB files

Place PDB files in data/raw/:

data/raw/2ito.pdb
data/raw/1A2C.pdb (example)
...

4. Add your input PDB files to config.yaml file

Update config.yaml with the PDB file names (without .pdb extension). Only include files that are actually in data/raw/:

pdb_files:
  - "2ito"
  - "1A2C"

Running the Workflow

1. Dry-run (optional)

snakemake -n

2. Execute the workflow

snakemake -j 1 #(or other number can be used)

(Use -j x to run x "jobs" in parallel)

Force re-run all steps

snakemake -F -j 1

Output Structure with two example PDB Files:

data/
├── processed/
│   ├── 2ito/
│   │   ├── 2ito_protein.pdb
│   │   ├── 2ito_ligand.pdb
│   │   ├── 2ito_protein.pdbqt
│   │   └── 2ito_ligand.pdbqt
│   └── 1A2C/
│       ├── 1A2C_protein.pdb
│       ├── 1A2C_ligand.pdb
│       ├── 1A2C_protein.pdbqt
│       └── 1A2C_ligand.pdbqt
└── docking/
    ├── 2ito/
    │   ├── 2ito_docked.sdf          # All poses that were in PDB File
    │   ├── 2ito_scores.csv          # Affinity scores
    │   ├── poses/
    │   │   ├── pose_00.sdf
    │   │   ├── pose_01.sdf
    │   │   └── ...
    │   └── visualize/
    │       ├── pose_00.png
    │       ├── pose_01.png
    │       └── ...
    └── 1A2C/
        ├── 1A2C_docked.sdf
        ├── 1A2C_scores.csv
        ├── poses/
        │   ├── pose_00.sdf
        │   ├── pose_01.sdf
        │   └── ...
        └── visualize/
            ├── pose_00.png
            ├── pose_01.png
            └── ...

Workflow Visualization

A DAG (Directed Acyclic Graph) visualization is included as dag_protein_docking.svg. This shows how the 6 rules are connected and their dependencies.

To regenerate the DAG after modifying the Snakefile:

snakemake --dag --snakefile Snakefile | dot -Tsvg > dag_protein_docking.svg

(Requires graphviz: conda install -c conda-forge graphviz)

File Structure

• Snakefile - Workflow definition with all 6 rules
• environment.yml - Conda dependencies specification
• config.yaml - Project configuration (paths, PDB files)
• scripts/ - Python scripts for each workflow step
  • prepare_structures.py - Extract protein/ligand from PDB
  • convert_to_pdbqt.py - Convert to PDBQT format
  • run_analysis.py - Extract docking scores
  • visualize_pose.py - Generate 3D visualizations

Citation

Based on TeachOpenCADD T015: Protein-Ligand Docking https://projects.volkamerlab.org/teachopencadd/talktorials/T015_protein_ligand_docking.html

PDB Files used in creation: https://www.rcsb.org/structure/2ITO https://www.rcsb.org/structure/2XNI

Contact

FHNW

Known limitations

• Unsupported atom types (e.g. Boron): The pipeline converts ligand and protein files to the PDBQT format and runs smina for docking. The AutoDock/Vina-style PDBQT format (used by smina) supports a limited set of atom types. Ligands that contain some elements (for example, boron B) may fail during PDBQT parsing and cause the workflow to stop with a parse error.

  Detection (quick): run this to list element symbols found in a ligand PDB:

  awk '{print substr($0,77,2)}' data/processed/<PDB>/<PDB>_ligand.pdb | sort | uniq -c

59d5283 (Initial project: Snakemake docking pipeline)