PRODIGY (PROtein binDIng enerGY prediction)

By default, PRODIGY treats each chain identifier as an isolated molecule. For complexes like antibody-antigen interactions, you can use the --selection option to group multiple chains into a single molecule or define specific pairs for contact calculation.

Syntax Rules:

• Use a space to separate chains that should be considered different molecules.
• Use a comma to include multiple chains as part of a single group.

Examples:

• --selection A B: Calculates contacts only between chains A and B.
• --selection A,B C: Calculates contacts between (A and C) and (B and C).
• --selection A B C: Calculates contacts between (A and B), (B and C), and (A and C).

# Example: Grouping chains A and B as one molecule, and C as another
prodigy --selection A,B C input.pdb

Install the prodigy-prot package using pip to access the binding affinity prediction CLI.

pip install prodigy-prot

PRODIGY uses poetry for dependency management and virtual environment handling. To set up a development environment, clone the repository and run poetry install to install all necessary dependencies.

git clone https://github.com/haddocking/prodigy.git && cd prodigy
poetry install

PRODIGY predicts binding affinity based on Intermolecular Contacts (ICs). You can run it on a single structure, a directory of structures, or a multi-model PDB file (ensemble).

Basic Usage Patterns:

• Single structure: prodigy <structure_file>
• Directory of molecules: prodigy <directory_with_molecules>
• Multi-model ensemble: prodigy <multi_model.pdb>

To speed up processing when running multiple structures, use the -np argument to specify the number of processors.

# Single structure
prodigy 3bzd.pdb

# Directory of structures with 2 processors
prodigy -np 2 input/

# Multi-model PDB
prodigy multi_model.pdb

Execute the project's test suite using the standard Python unittest module.

python -m unittest

PRODIGY provides different levels of output detail:

• --quiet (-q): Minimizes output to only the predicted affinity value. This is useful for programmatic parsing.
• --showall (-s): Outputs all original PRODIGY features, but excludes the Buried Surface Area (BSA).

Note: --quiet and --showall are mutually exclusive and cannot be used together.

By default, PRODIGY treats each chain as an isolated molecule. For complexes like antibody-antigen interfaces, you may want to group multiple chains into a single molecule. Use the --selection flag to define these groups.

• Use spaces to separate different molecules.
• Use commas to group chains into the same molecule.

Examples:

• --selection A B: Calculates contacts only between chain A and chain B.
• --selection A,B C: Calculates contacts between (A and C) and (B and C). Chains A and B are treated as one molecule.
• --selection A B C: Calculates contacts between A&B, B&C, and A&C.

When running PRODIGY on multiple structures and intending to parse the results programmatically, use the -q or --quiet flag. This suppresses the detailed feature breakdown and outputs only the predicted affinity value per file.

$ prodigy -q -np 2 input/
3bzd  -9.373
2oob  -6.230
1ppe  -14.727

The print_prediction() method outputs the prediction results in a human-readable format. You can direct the output to a file or stdout.

Arguments:

• outfile (str): Path to a file where the report will be written. If omitted, it prints to stdout.
• quiet (bool): If True, only prints the name and the binding affinity (ba_val).
• showall (bool): If True, includes additional details like hydrophobic contact counts (LL, BL, BB) and NIS percentages.

# Standard detailed output to stdout
prodigy.print_prediction()

# Quiet output to a file
prodigy.print_prediction(outfile="results.txt", quiet=True)

# Full detailed output including hydrophobic counts
prodigy.print_prediction(showall=True)

The as_dict() method returns a dictionary containing all calculated values for the prediction. This is useful for programmatic access or saving results to JSON.

Keys included in the dictionary:

• model: The ID of the Biopython model.
• selection: The chain selection used.
• temp: The temperature used (in Celsius).
• ICs: Number of intermolecular contacts.
• nis_a: Percentage of apolar NIS residues.
• nis_c: Percentage of charged NIS residues.
• nis_p: Percentage of polar NIS residues.
• ba_val: Predicted binding affinity (kcal/mol).
• kd_val: Predicted dissociation constant (M).
• CC, CP, AC, PP, AP, AA, LL, BL, BB: Counts of specific contact types (e.g., Charged-Charged, Polar-Polar, etc.).

results = prodigy.as_dict()
# Access specific metrics
print(results['ba_val'])
print(results['CC'])