Bunsalyze

For questions/issues/feedback please use GitHub Issues or contact Benjamin Fry (bfry@g.harvard.edu)

---

Description

Python module for counting buried polar atoms with unsatisfied hydrogen bonds. Used for ranking and selection of designed Protein-Ligand interfaces generated by LASErMPNN / NISE. The lower the number of buried unsatisfied polar atoms, the better the design. We typically score protein-ligand interfaces by upweighting the ligand BUN contributions by 2x relative to the protein BUN contributions.

Installation

Runs with the LASErMPNN conda environment.

Alternatively, install with pip install bunsalyze and run with bunsalyze ....

Usage

From the directory containing this github repository and run:

python -m bunsalyze ./path/to/pdb_file.pdb 'smiles_string_for_ligand'

Alternatively, install with pip install bunsalyze and run with bunsalyze ....

Here is the result of the --help flag:

usage: bunsalyze.py [-h] [--sasa_threshold SASA_THRESHOLD] [--alpha_hull_alpha ALPHA_HULL_ALPHA] [--output OUTPUT] [--disable_hydrogen_clash_check] [--override_ligand_selection_string OVERRIDE_LIGAND_SELECTION_STRING] [--ncaa_dict NCAA_DICT] [--ignore_sulfur_acceptors] [--ignore_sasa_threshold] input_path smiles

Analyze protein-ligand complexes for buried unsatisfied polar atoms (BUNs).

positional arguments:
  input_path            Path to the PDB file containing the protein-ligand complex
  smiles                SMILES string representing the ligand

options:
  -h, --help            show this help message and exit
  --sasa_threshold SASA_THRESHOLD
                        SASA threshold for burial (default: 1.0 Ų)
  --alpha_hull_alpha ALPHA_HULL_ALPHA
                        Convex hull alpha parameter, default is 9.0
  --output OUTPUT       Output file path (default: print to stdout)
  --disable_hydrogen_clash_check
                        Default behavior doesn't count hbonds made at the expense of a hydrogen vdW clash. Set this flag to disable that check.
  --override_ligand_selection_string OVERRIDE_LIGAND_SELECTION_STRING
                        How to select the ligand from the PDB file, default is "not protein" but this fails with noncanonical amino acids.
  --ncaa_dict NCAA_DICT
                        Dictionary mapping ncaa 3-letter code to polar atoms which map to tuples of (# hbonds atom can accept, list of atom names of attached donor hydrogens). Format: '{"DJD": {"N": (0, ["H"]), "O": (2, []), "N03": (1, []), "N04": (1, []), "N05": (1, []), "N06": (1, [])}}'
  --ignore_sulfur_acceptors
                        If set, ignores sulfur atoms as potential acceptors. Default behavior includes sulfur atoms as acceptors.
  --ignore_sasa_threshold
                        If set, does not use a SASA threshold to determine burial, only uses convex hull. Default behavior uses both SASA and convex hull.

Here is a result of running the bunsalyze function on the PDB found in example_pdbs. The "BUNS Score" used in NISE is the sum of the lengths of the ligand buns (x2) and protein buns lists.

$ python -m bunsalyze.bunsalyze ./bunsalyze/example_pdbs/epic_1.pdb 'CC[C@]1(O)C2=C(C(N3CC4=C5[C@@H]([NH3+])CCC6=C5C(N=C4C3=C2)=CC(F)=C6C)=O)COC1=O'
Ligand Atom SASA: (< 1.0 A^2 == buried)
        O 4.069010588001133
        N2 16.1801952763787
        O2 0.0
        N3 0.0
        O3 0.0
        O4 0.0
{'input_path': './bunsalyze/example_pdbs/epic_1.pdb',
 'ligand_atoms_buried_sasa': ['O2', 'N3', 'O3', 'O4'],
 'ligand_atoms_in_hull': ['O', 'O2', 'N3', 'O3', 'O4'],
 'ligand_atoms_sasa': {'N2': 16.1801952763787,
                       'N3': 0.0,
                       'O': 4.069010588001133,
                       'O2': 0.0,
                       'O3': 0.0,
                       'O4': 0.0},
 'ligand_buns': [('N3', np.str_('X'), np.str_('X9E'), 1, np.str_('')),
                 ('O3', np.str_('X'), np.str_('X9E'), 1, np.str_('')),
                 ('O4', np.str_('X'), np.str_('X9E'), 1, np.str_(''))],
 'ligand_buried_fraction_unsat': {(np.str_('X'), np.str_('X9E'), 1, np.str_('')): 0.875},
 'ligand_fraction_unsat': {(np.str_('X'), np.str_('X9E'), 1, np.str_('')): 0.8461538461538461},
 'protein_buns': [(np.str_('OE1'),
                   np.str_('A'),
                   np.str_('GLU'),
                   1,
                   np.str_('')),
                  (np.str_('NE2'),
                   np.str_('A'),
                   np.str_('GLN'),
                   51,
                   np.str_(''))],
 'protein_buried_fraction_unsat': {(np.str_('A'), np.str_('GLN'), 51, np.str_('')): 0.75,
                                   (np.str_('A'), np.str_('GLU'), 1, np.str_('')): 1.0},
 'protein_fraction_unsat': {(np.str_('A'), np.str_('GLN'), 51, np.str_('')): 0.75,
                            (np.str_('A'), np.str_('GLU'), 1, np.str_('')): 1.0}}

TODO:

• Add support for multiple ligands
• Add support for Ca-H ... Acc hbonds