CovCysPredictor

CovCysPredictor is a tool that is for predicting the ligandability (=reactivity + selectivity) of a particular cysteine residue in a protein. It takes as input structural information in the form of a prepared PDB file, and outputs a prediction file with one entry for each cysteine giving relative ligandability predictions. The predictions are based on an interpretable model including the solvent exposure, the inclusion in a pocket, and the local amino acid environment.

Using CovCysPredictor

Dependencies

This tool has the following dependencies:

• Python (3.10+)
• conda
• fpocket for pocket prediction

The relevant Python packages (Biopython, etc) are conveniently available in a conda environment.yml file, but otherwise Python package dependencies can be managed manually.

The location of your fpocket directory may need to be updated in run_cysteine_prediction.sh.

This tool can be run using the following command:

./run_cysteine_prediction.sh pdbs/1a55_edited.pdb example_outputs

And will issue results something like this (ex for PDBID 1a54):

{
    'A 197': {
        'chain': 'A',
        'resid': 197,
        'sasa': 19.301945263655696,
        'log_exp': 3.0107167072525907,
        'any_fpocket': 1,
        'neighbors': ['P', 'A', 'P', 'V', 'E', 'Y', 'Y', 'A', 'K', 'Q', 'L', 'D'],
        'score': '0.32',
        'predicted_modifiable': True
    }
}