pocket-eigenvalues

Python script for extracting atomic coordinates from PDB pocket structures and computing eigenvalue-based geometric descriptors (covariance eigenvalues and ratios) for RNA & protein binding pockets.

REQUIREMENTS

• Python ≥ 3.8
• Biopython
• NumPy
• Pandas

Dependencies can be installed using:

pip install -r requirements.txt

INPUT

• A directory containing pocket PDB file(s) (one pocket per file).
• Each pocket is identified by its filename.

USAGE

python pocket_xyz_eigenvalues.py data/ results.csv

Where:

• <input_folder> is the path to a directory containing pocket PDB file(s) (e.g. data/)
• <output_csv> is the name of the CSV file to be generated (e.g. results.csv)

OUTPUT

The script produces a CSV file containing the following columns:

• PocketID: pocket identifier (filename without .pdb)
• Eigval1, Eigval2, Eigval3: eigenvalues of the coordinate covariance matrix
• rEV32, rEV31, rEV21: eigenvalue ratios
• NumAtoms: number of atoms in the pocket
• Rg: radius of gyration of the pocket

EXAMPLE DATA

The data/ directory contains two example pocket PDB files provided for testing the pipeline.