Common Subgraph Detection in Protein Structures

This project identifies common subgraphs in protein structures using the Graphein library. It can be used with any protein structures, mapping common subgraphs to highlight areas of structural similarity between them.

Overview

The algorithm is inspired by work on protein-protein binding sites and surface structure conservation. It analyzes the input protein structures to detect common subgraphs and map them onto the structural representation of the proteins.

Requirements

• Python: Tested with Python 3.8
• Graphein: Make sure Graphein is installed in a Conda environment.

Known Issues

• Graphein's centroid computation issue with multiple chains was resolved by modifying the file at: graphein/protein/graphs.py

  Replace:

  ["residue_number", "chain_id", "residue_name", "insertion"]

  with:

  ["chain_id", "residue_number", "residue_name", "insertion"]

• Compatibility issues when running DSSP with Graphein were resolved by modifying the file at: graphein/protein/features/nodes/dssp.py

  Change:

  pdb_file, DSSP=executable, dssp_version=dssp_version

  to:

  pdb_file, DSSP=executable

Installation

1. Clone the repository:

   git clone https://github.com/LBC-LNBio/pMHC_graph.git

2. Create a Conda environment and install the required packages:

   conda create -n graphein_env python=3.8
   conda activate graphein_env
   pip install -r requirements.txt

3. Install Graphein:

   pip install graphein

Usage

Command Example

Run the script with the following command:

python3 main.py --mols_path /path/to/pdb_folder --residues_lists /path/to/residues_list.json --centroid_threshold 10 --run_name test_run --association_mode similarity --output_path ./output_directory

Arguments

Argument	Description	Default
--mols_path	Path with PDB input files.	''
--centroid_threshold	Distance threshold for building the interface graphs.	10
--run_name	Name for storing results in the output folder.	test
--association_mode	Mode for creating association nodes (identity/similarity).	identity
--output_path	Path to store output results.	'~/'
--neighbor_similarity_cutoff	Threshold for neighbor's similarity.	0.95
--rsa_filter	Threshold for filtering residues by RSA.	0.1
--rsa_similarity_threshold	Threshold for RSA similarity in association graphs.	0.95
--residues_lists	Path to JSON file containing the pdb residues.	None
--debug	Activate debug mode.	False

Output

• Common Subgraphs: The script generates common subgraphs that are mapped to the protein structures.
• Visualization: Graphs of the common subgraphs are saved as images in the output directory.

Logging

The script supports debug mode for detailed logging. Activate debug mode by setting the --debug flag to True.

References

• Original Algorithm: The algorithm is inspired by methods used in protein-protein binding sites and surface structure analysis.

License

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

Acknowledgements

• Graphein Library: Used for protein graph analysis.