♻️ add typing

  - bash scripts/install_deepchem_conda.sh deepchem

  - conda activate deepchem

  - python setup.py install

  - pip install coveralls mypy yapf==0.22.0

  - pip install coveralls mypy flake8 yapf==0.22.0

script:

  - bash devtools/run_yapf.sh

"""

import logging

import numpy as np

from typing import Any, Optional, Tuple

from deepchem.models import Model

from deepchem.utils import rdkit_util

from deepchem.utils import coordinate_box_utils as box_utils

from deepchem.utils.fragment_util import get_contact_atom_indices

logger = logging.getLogger(__name__)

def extract_active_site(protein_file, ligand_file, cutoff=4):

def extract_active_site(protein_file: str,

                        ligand_file: str,

                        cutoff: float = 4.0

                       ) -> Tuple[box_utils.CoordinateBox, np.ndarray]:

  """Extracts a box for the active site.

  Parameters

    Location of protein PDB

  ligand_file: str

    Location of ligand input file

  cutoff: int, optional

  cutoff: float, optional (default 4.0)

    The distance in angstroms from the protein pocket to

    consider for featurization.

  technique to be used.

  """

  def find_pockets(self, molecule):

  def find_pockets(self, molecule: Any):

    """Finds potential binding pockets in proteins.

    Parameters

  Based on https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4112621/pdf/1472-6807-14-18.pdf

  """

  def __init__(self, scoring_model=None, pad=5):

  def __init__(self, scoring_model: Optional[Model] = None, pad: int = 5):

    """Initialize the pocket finder.

    Parameters

    ----------

    scoring_model: `dc.models.Model`, optional

      If specified, use this model to prune pockets.

    pad: float, optional

    pad: int, optional (default 5)

      The number of angstroms to pad around a binding pocket's atoms

      to get a binding pocket box.

    """

    self.scoring_model = scoring_model

    self.pad = pad

  def find_all_pockets(self, protein_file):

  def find_all_pockets(self, protein_file: str):

    """Find list of binding pockets on protein.

    Parameters

    coords, _ = rdkit_util.load_molecule(protein_file)

    return box_utils.get_face_boxes(coords, self.pad)

  def find_pockets(self, macromolecule_file):

  def find_pockets(self, macromolecule_file: str):

    """Find list of suitable binding pockets on protein.

    This function computes putative binding pockets on this protein.

"""

import logging

import tempfile

from typing import Any, Optional, cast

from deepchem.models import Model

from deepchem.feat import ComplexFeaturizer

from deepchem.data import NumpyDataset

from deepchem.dock import PoseGenerator

logger = logging.getLogger(__name__)

  generation and scoring classes that are provided to this class.

  """

  def __init__(self, pose_generator, featurizer=None, scoring_model=None):

  def __init__(self,

               pose_generator: PoseGenerator,

               featurizer: Optional[ComplexFeaturizer] = None,

               scoring_model: Optional[Model] = None):

    """Builds model.

    Parameters

    ----------

    pose_generator: `PoseGenerator`

      The pose generator to use for this model

    featurizer: `ComplexFeaturizer`

    featurizer: `ComplexFeaturizer`, optional (default None)

      Featurizer associated with `scoring_model`

    scoring_model: `Model`

    scoring_model: `Model`, optional (default None)

      Should make predictions on molecular complex.

    """

    if ((featurizer is not None and scoring_model is None) or

    self.scoring_model = scoring_model

  def dock(self,

           molecular_complex,

           centroid=None,

           box_dims=None,

           exhaustiveness=10,

           num_modes=9,

           num_pockets=None,

           out_dir=None,

           use_pose_generator_scores=False):

           molecular_complex: Any,

           centroid: Optional[int] = None,

           box_dims: Optional[int] = None,

           exhaustiveness: int = 10,

           num_modes: int = 9,

           num_pockets: Optional[int] = None,

           out_dir: Optional[str] = None,

           use_pose_generator_scores: bool = False):

    """Generic docking function.

    This docking function uses this object's featurizer, pose

      raise ValueError(

          "Cannot set use_pose_generator_scores=True when self.scoring_model is set (since both generator scores for complexes)."

      )

    outputs = self.pose_generator.generate_poses(

        molecular_complex,

        centroid=centroid,

      complexes, scores = outputs

    else:

      complexes = outputs

    # We know use_pose_generator_scores == False in this case

    if self.scoring_model is not None:

      for posed_complex in complexes:

        # NOTE: this casting is workaround. This line doesn't effect anything to the runtime

        self.featurizer = cast(ComplexFeaturizer, self.featurizer)

        # TODO: How to handle the failure here?

        features, _ = self.featurizer.featurize([molecular_complex])

        features, _ = self.featurizer.featurize(  # type: ignore

            [molecular_complex])

        dataset = NumpyDataset(X=features)

        score = self.scoring_model.predict(dataset)

        yield (posed_complex, score)

import os

import tempfile

import tarfile

import numpy as np

from subprocess import call

from subprocess import check_output

from typing import Optional, Tuple

from deepchem.dock.binding_pocket import BindingPocketFinder

from deepchem.utils import rdkit_util

from deepchem.utils import mol_xyz_util

from deepchem.utils import geometry_utils

  """

  def generate_poses(self,

                     molecular_complex,

                     centroid=None,

                     box_dims=None,

                     exhaustiveness=10,

                     num_modes=9,

                     num_pockets=None,

                     out_dir=None,

                     generate_scores=False):

                     molecular_complex: Tuple[str, str],

                     centroid: Optional[np.ndarray] = None,

                     box_dims: Optional[np.ndarray] = None,

                     exhaustiveness: int = 10,

                     num_modes: int = 9,

                     num_pockets: Optional[int] = None,

                     out_dir: Optional[str] = None,

                     generate_scores: bool = False):

    """Generates a list of low energy poses for molecular complex

    Parameters

    ----------

    molecular_complexes: list

      A representation of a molecular complex.

    centroid: np.ndarray, optional

    molecular_complexes: Tuple[str, str]

      A representation of a molecular complex. This is a tuple of

      (protein_file, ligand_file).

    centroid: np.ndarray, optional (default None)

      The centroid to dock against. Is computed if not specified.

    box_dims: np.ndarray, optional

    box_dims: np.ndarray, optional (default None)

      Of shape `(3,)` holding the size of the box to dock. If not

      specified is set to size of molecular complex plus 5 angstroms.

    exhaustiveness: int, optional (default 10)

      If specified, `self.pocket_finder` must be set. Will only

      generate poses for the first `num_pockets` returned by

      `self.pocket_finder`.

    out_dir: str, optional

    out_dir: str, optional (default None)

      If specified, write generated poses to this directory.

    generate_score: bool, optional (default False)

      If `True`, the pose generator will return scores for complexes.

  This class requires RDKit to be installed.

  """

  def __init__(self, sixty_four_bits=True, pocket_finder=None):

  def __init__(self,

               sixty_four_bits: bool = True,

               pocket_finder: Optional[BindingPocketFinder] = None):

    """Initializes Vina Pose Generator

    Parameters

      os.remove(downloaded_file)

  def generate_poses(self,

                     molecular_complex,

                     centroid=None,

                     box_dims=None,

                     exhaustiveness=10,

                     num_modes=9,

                     num_pockets=None,

                     out_dir=None,

                     generate_scores=False):

                     molecular_complex: Tuple[str, str],

                     centroid: Optional[np.ndarray] = None,

                     box_dims: Optional[np.ndarray] = None,

                     exhaustiveness: int = 10,

                     num_modes: int = 9,

                     num_pockets: Optional[int] = None,

                     out_dir: Optional[str] = None,

                     generate_scores: bool = False):

    """Generates the docked complex and outputs files for docked complex.

    TODO: How can this work on Windows? We need to install a .msi file and invoke it correctly from Python for this to work.

    Parameters

    ----------

    molecular_complexes: list

      A representation of a molecular complex.

    molecular_complexes: Tuple[str]

      A representation of a molecular complex. This is a tuple of

      (protein_file, ligand_file).

    centroid: np.ndarray, optional

      The centroid to dock against. Is computed if not specified.

    box_dims: np.ndarray, optional

      else:

        # I'm not sure why specifying the args as a list fails on other platforms,

        # but for some reason it only works if I pass it as a string.

        args = "%s --config %s --log %s --out %s" % (self.vina_cmd, conf_file,

                                                     log_file, out_pdbqt)

        args = "%s --config %s --log %s --out %s" % (  # type: ignore

            self.vina_cmd, conf_file, log_file, out_pdbqt)

      # FIXME: We should use `subprocess.run` instead of `call`

      call(args, shell=True)

      ligands, scores = vina_utils.load_docked_ligands(out_pdbqt)

      docked_complexes += [(protein_mol[1], ligand) for ligand in ligands]

import numpy as np

def pairwise_distances(coords1, coords2):

def pairwise_distances(coords1: np.ndarray, coords2: np.ndarray) -> np.ndarray:

  """Returns matrix of pairwise Euclidean distances.

  Parameters

  Returns

  -------

  np.ndarray

    A `(N,M)` array with pairwise distances.

  """

  return np.sum((coords1[None, :] - coords2[:, None])**2, -1)**0.5

def cutoff_filter(d, x, cutoff=8.0):

def cutoff_filter(d: np.ndarray, x: np.ndarray, cutoff=8.0) -> np.ndarray:

  """Applies a cutoff filter on pairwise distances

  Parameters

  Returns

  -------

  A `(N,M)` array with values where distance is too large thresholded

  to 0.

  np.ndarray

    A `(N,M)` array with values where distance is too large thresholded to 0.

  """

  return np.where(d < cutoff, x, np.zeros_like(x))

def vina_nonlinearity(c, w, Nrot):

def vina_nonlinearity(c: np.ndarray, w: float, Nrot: int) -> np.ndarray:

  """Computes non-linearity used in Vina.

  Parameters

  Returns

  -------

  np.ndarray

    A `(N, M)` array with activations under a nonlinearity.

  """

  out_tensor = c / (1 + w * Nrot)

  return out_tensor

def vina_repulsion(d):

def vina_repulsion(d: np.ndarray) -> np.ndarray:

  """Computes Autodock Vina's repulsion interaction term.

  Parameters

  Returns

  -------

  np.ndarray

    A `(N, M)` array with repulsion terms.

  """

  return np.where(d < 0, d**2, np.zeros_like(d))

def vina_hydrophobic(d):

def vina_hydrophobic(d: np.ndarray) -> np.ndarray:

  """Computes Autodock Vina's hydrophobic interaction term.

  Here, d is the set of surface distances as defined in:

  Jain, Ajay N. "Scoring noncovalent protein-ligand interactions: a continuous differentiable function tuned to compute binding affinities." Journal of computer-aided molecular design 10.5 (1996): 427-440.

  Here, d is the set of surface distances as defined in [1]_

  Parameters

  ----------

  Returns

  -------

  A `(N, M)` array of hydrophoboic interactions in a piecewise linear

  curve.

  np.ndarray

    A `(N, M)` array of hydrophoboic interactions in a piecewise linear curve.

  References

  ----------

  .. [1] Jain, Ajay N. "Scoring noncovalent protein-ligand interactions:

     a continuous differentiable function tuned to compute binding affinities."

     Journal of computer-aided molecular design 10.5 (1996): 427-440.

  """

  out_tensor = np.where(d < 0.5, np.ones_like(d),

                        np.where(d < 1.5, 1.5 - d, np.zeros_like(d)))

  return out_tensor

def vina_hbond(d):

def vina_hbond(d: np.ndarray) -> np.ndarray:

  """Computes Autodock Vina's hydrogen bond interaction term.

  Here, d is the set of surface distances as defined in:

  Jain, Ajay N. "Scoring noncovalent protein-ligand interactions: a continuous differentiable function tuned to compute binding affinities." Journal of computer-aided molecular design 10.5 (1996): 427-440.

  Here, d is the set of surface distances as defined in [1]_

  Parameters

  ----------

  Returns

  -------

  A `(N, M)` array of hydrophoboic interactions in a piecewise linear

  curve.

  np.ndarray

    A `(N, M)` array of hydrophoboic interactions in a piecewise linear curve.

  References

  ----------

  .. [1] Jain, Ajay N. "Scoring noncovalent protein-ligand interactions:

     a continuous differentiable function tuned to compute binding affinities."

     Journal of computer-aided molecular design 10.5 (1996): 427-440.

  """

  out_tensor = np.where(

      d < -0.7, np.ones_like(d),

  return out_tensor

def vina_gaussian_first(d):

def vina_gaussian_first(d: np.ndarray) -> np.ndarray:

  """Computes Autodock Vina's first Gaussian interaction term.

  Here, d is the set of surface distances as defined in [1]_

  Returns

  -------

  np.ndarray

    A `(N, M)` array of gaussian interaction terms.

  References

  return out_tensor

def vina_gaussian_second(d):

def vina_gaussian_second(d: np.ndarray) -> np.ndarray:

  """Computes Autodock Vina's second Gaussian interaction term.

  Here, d is the set of surface distances as defined in [1]_

  Returns

  -------

  np.ndarray

    A `(N, M)` array of gaussian interaction terms.

  References

  return out_tensor

def weighted_linear_sum(w, x):

def weighted_linear_sum(w: np.ndarray, x: np.ndarray) -> np.ndarray:

  """Computes weighted linear sum.

  Parameters

    Of shape `(N,)`

  x: np.ndarray

    Of shape `(N,)`

  Returns

  -------

  np.ndarray

    A scalar value

  """

  return np.sum(np.dot(w, x))

def vina_energy_term(coords1, coords2, weights, wrot, Nrot):

def vina_energy_term(coords1: np.ndarray, coords2: np.ndarray,

                     weights: np.ndarray, wrot: float, Nrot: int) -> np.ndarray:

  """Computes the Vina Energy function for two molecular conformations

  Parameters

  Returns

  -------

  Scalar with energy

  np.ndarray

    A scalar value with free energy

  """

  # TODO(rbharath): The autodock vina source computes surface distances which take into account the van der Waals radius of each atom type.

  dists = pairwise_distances(coords1, coords2)