Trying to hbond featurizers working (2f144e31) · Commits · 钟慕尧 / deepchem

deepchem/feat/contact_fingerprints.py

+13 −1

Original line number	Diff line number	Diff line
		@@ -54,7 +54,15 @@ def featurize_binding_pocket_ecfp(protein_xyz,
		class ContactCircularFingerprint(ComplexFeaturizer):
		"""Compute (Morgan) fingerprints near contact points of macromolecular complexes.

		Given a macromolecular complex made up of multiple constituent molecules, first compute the contact points where atoms from different molecules come close to one another. For atoms within "contact regions," compute radial "ECFP" fragments which are sub-molecules centered at atoms in the contact region.
		Given a macromolecular complex made up of multiple
		constituent molecules, first compute the contact points where
		atoms from different molecules come close to one another. For
		atoms within "contact regions," compute radial "ECFP"
		fragments which are sub-molecules centered at atoms in the
		contact region.

		For a macromolecular complex, returns a vector of shape
		`(2*size,)`
		"""

		def __init__(self,
		@@ -120,6 +128,10 @@ class ContactCircularVoxelizer(ComplexFeaturizer):
		fragments which are sub-molecules centered at atoms in the
		contact region. Localize these ECFP fingeprints at the voxel
		in which they originated.

		Featurizes a macromolecular complex into a tensor of shape
		`(voxels_per_edge, voxels_per_edge, voxels_per_edge, size)`
		where `voxels_per_edge = int(box_width/voxel_width)`.
		"""

		def __init__(self,

deepchem/feat/grid_featurizers.py

+101 −22

Original line number	Diff line number	Diff line
		@@ -4,6 +4,8 @@ Compute various spatial fingerprints for macromolecular complexes.
		import logging
		from deepchem.utils.rdkit_util import get_partial_charge
		from deepchem.feat import ComplexFeaturizer
		from deepchem.utils.hash_utils import hash_ecfp_pair
		from deepchem.utils.hash_utils import vectorize
		from deepchem.utils.voxel_utils import voxelize
		from deepchem.utils.rdkit_util import compute_salt_bridges
		from deepchem.utils.rdkit_util import compute_binding_pocket_cation_pi
		@@ -11,9 +13,14 @@ from deepchem.utils.voxel_utils import convert_atom_to_voxel
		from deepchem.utils.voxel_utils import convert_atom_pair_to_voxel
		from deepchem.utils.rdkit_util import compute_pairwise_distances
		from deepchem.utils.rdkit_util import compute_pi_stack
		from deepchem.utils.rdkit_util import compute_hydrogen_bonds

		logger = logging.getLogger(__name__)


		HBOND_DIST_BINS = [(2.2, 2.5), (2.5, 3.2), (3.2, 4.0)]
		HBOND_ANGLE_CUTOFFS = [5, 50, 90]

		def compute_charge_dictionary(molecule):
		"""Create a dictionary with partial charges for each atom in the molecule.

		@@ -37,9 +44,9 @@ class ChargeVoxelizer(ComplexFeaturizer):
		a local charge array. Sum contributions to get an effective
		charge at each voxel.

		Creates a tensor output of shape (box_width, box_width,
		box_width, 1) for each macromolecular complex that computes
		the effective charge at each voxel.
		Creates a tensor output of shape `(voxels_per_edge,
		voxels_per_edge, voxels_per_edge, 1)` for each macromolecular
		complex that computes the effective charge at each voxel.
		"""
		def __init__(self,
		box_width=16.0,
		@@ -96,8 +103,8 @@ class SaltBridgeVoxelizer(ComplexFeaturizer):
		interact in a salt-bridge, the interaction is double counted
		in both voxels.

		Creates a tensor output of shape (box_width, box_width,
		box_width, 1) for each macromolecular complex that computes
		Creates a tensor output of shape `(voxels_per_edge,
		voxels_per_edge, voxels_per_edge, 1)` for each macromolecular
		the number of salt bridges at each voxel.
		"""
		def __init__(self,
		@@ -158,8 +165,8 @@ class CationPiVoxelizer(ComplexFeaturizer):
		bridge in the voxel in which it originated to create a local
		cation-pi array.

		Creates a tensor output of shape (box_width, box_width,
		box_width, 1) for each macromolecular complex that computes
		Creates a tensor output of shape `(voxels_per_edge,
		voxels_per_edge, voxels_per_edge, 1)` for each macromolecular
		the number of cation-pi interactions at each voxel.
		"""
		def __init__(self,
		@@ -232,11 +239,11 @@ class PiStackVoxelizer(ComplexFeaturizer):
		localize this salt bridge in the voxel in which it originated
		to create a local pi-stacking array.

		Creates a tensor output of shape (box_width, box_width,
		box_width, 2) for each macromolecular complex. Each voxel has
		2 fields, with the first tracking the number of pi-pi
		parallel interactions, and the second tracking the number of
		pi-T interactions.
		Creates a tensor output of shape `(voxels_per_edge,
		voxels_per_edge, voxels_per_edge, 2)` for each macromolecular
		Each voxel has 2 fields, with the first tracking the number
		of pi-pi parallel interactions, and the second tracking the
		number of pi-T interactions.
		"""
		def __init__(self,
		distance_cutoff=4.4,
		@@ -324,6 +331,75 @@ class PiStackVoxelizer(ComplexFeaturizer):
		nb_channel=1)
		return [pi_parallel_tensor, pi_t_tensor]

		class HydrogenBondCounter(ComplexFeaturizer):
		"""Counts hydrogen bonds between atoms in macromolecular complexes.

		Given a macromolecular complex made up of multiple
		constitutent molecules, count the number hydrogen bonds
		between atoms in the macromolecular complex.

		Creates a scalar output of shape `(1,)` for each
		macromolecular that computes the total number of hydrogen
		bonds.
		"""
		def __init__(self,
		distance_bins=None,
		angle_cutoffs=None):
		"""
		Parameters
		----------
		distance_bins: list[tuple]
		List of hydgrogen bond distance bins. If not specified is
		set to default
		`[(2.2, 2.5), (2.5, 3.2), (3.2, 4.0)]`.
		angle_cutoffs: list[float]
		List of hydrogen bond angle cutoffs. Max allowed
		deviation from the ideal (180 deg) angle between
		hydrogen-atom1, hydrogen-atom2 vectors.If not specified
		is set to default `[5, 50, 90]`
		"""
		if distance_bins is None:
		self.distance_bins = HBOND_DIST_BINS
		else:
		self.distance_bins = distance_bins
		if angle_cutoffs is None:
		self.angle_cutoffs = HBOND_ANGLE_CUTOFFS
		else:
		self.angle_cutoffs = angle_cutoffs
		######################################################
		print("constructor")
		print("self.distance_bins")
		print(self.distance_bins)
		print("self.angle_cutoffs")
		print(self.angle_cutoffs)
		######################################################

		def _featurize_complex(self, mol, protein):
		"""
		Compute featurization for a single mol/protein complex

		Parameters
		----------
		mol: object
		Representation of the molecule
		protein: object
		Representation of the protein
		"""
		(lig_xyz, lig_rdk), (prot_xyz, prot_rdk) = mol, protein
		distances = compute_pairwise_distances(prot_xyz, lig_xyz)
		######################################################
		print("self.distance_bins")
		print(self.distance_bins)
		print("self.angle_cutoffs")
		print(self.angle_cutoffs)
		######################################################
		return [
		vectorize(
		hash_ecfp_pair, feature_list=hbond_list, size=1)
		for hbond_list in compute_hydrogen_bonds(
		prot_xyz, prot_rdk, lig_xyz, lig_rdk, distances, self.distance_bins, self.angle_cutoffs)
		]

		class HydrogenBondVoxelizer(ComplexFeaturizer):
		"""Localize hydrogen bonds between atoms in macromolecular complexes.

		@@ -335,8 +411,8 @@ class HydrogenBondVoxelizer(ComplexFeaturizer):
		different voxels interact in a hydrogen bond, the interaction
		is double counted in both voxels.

		Creates a tensor output of shape (box_width, box_width,
		box_width, 1) for each macromolecular complex that computes
		Creates a tensor output of shape `(voxels_per_edge,
		voxels_per_edge, voxels_per_edge, 1)` for each macromolecular
		the number of hydrogen bonds at each voxel.
		"""
		def __init__(self,
		@@ -347,13 +423,15 @@ class HydrogenBondVoxelizer(ComplexFeaturizer):
		"""
		Parameters
		----------
		distance_cutoff: float, optional (default 4.0)
		The distance in angstroms within which atoms must be to
		be considered for a hydrogen bond interaction between
		them.
		angle_cutoff: float, optional (default 40.0)
		Angle cutoff. Max allowed deviation from the ideal (180
		deg) angle between hydrogen-atom1, hydrogen-atom2 vectors.
		distance_bins: list[tuple]
		List of hydgrogen bond distance bins. If not specified is
		set to default
		`[(2.2, 2.5), (2.5, 3.2), (3.2, 4.0)]`.
		angle_cutoffs: list[float]
		List of hydrogen bond angle cutoffs. Max allowed
		deviation from the ideal (180 deg) angle between
		hydrogen-atom1, hydrogen-atom2 vectors.If not specified
		is set to default `[5, 50, 90]`
		box_width: float, optional (default 16.0)
		Size of a box in which voxel features are calculated. Box
		is centered on a ligand centroid.
		@@ -361,6 +439,7 @@ class HydrogenBondVoxelizer(ComplexFeaturizer):
		Size of a 3D voxel in a grid.
		"""
		self.distance_cutoff = distance_cutoff
		self.angle_cutoff = angle_cutoff
		self.box_width = box_width
		self.voxel_width = voxel_width
		self.voxels_per_edge = int(self.box_width / self.voxel_width)

deepchem/feat/rdkit_grid_featurizer.py

+6 −7

Original line number	Diff line number	Diff line
		@@ -21,6 +21,7 @@ from deepchem.feat.grid_featurizers import ChargeVoxelizer
		from deepchem.feat.grid_featurizers import SaltBridgeVoxelizer
		from deepchem.feat.grid_featurizers import CationPiVoxelizer
		from deepchem.feat.grid_featurizers import PiStackVoxelizer
		from deepchem.feat.grid_featurizers import HydrogenBondCounter
		from deepchem.feat.grid_featurizers import HydrogenBondVoxelizer
		from deepchem.utils.rdkit_util import compute_hydrogen_bonds
		from deepchem.utils.rdkit_util import load_molecule
		@@ -213,6 +214,9 @@ class RdkitGridFeaturizer(ComplexFeaturizer):
		angle_cutoff=self.cutoffs['pi_stack_angle_cutoff'],
		box_width=self.box_width,
		voxel_width=self.voxel_width)
		self.hbond_counter = HydrogenBondCounter(
		distance_bins=self.cutoffs['hbond_dist_cutoff'],
		angle_cutoffs=self.cutoffs['hbond_angle_cutoff'])
		self.hbond_voxelizer = HydrogenBondVoxelizer(
		distance_cutoff=self.cutoffs['hbond_dist_cutoff'],
		angle_cutoff=self.cutoffs['hbond_angle_cutoff'],
		@@ -272,12 +276,8 @@ class RdkitGridFeaturizer(ComplexFeaturizer):
		return self.splif_featurizer._featurize_complex((lig_xyz, lig_rdk),
		(prot_xyz, prot_rdk))
		if feature_name == 'hbond_count':
		return [
		vectorize(hash_ecfp_pair, feature_list=hbond_list, size=1)
		for hbond_list in compute_hydrogen_bonds(
		prot_xyz, prot_rdk, lig_xyz, lig_rdk, distances, self.cutoffs[
		'hbond_dist_bins'], self.cutoffs['hbond_angle_cutoffs'])
		]
		return self.hbond_counter._featurize_complex((lig_xyz, lig_rdk),
		(prot_xyz, prot_rdk))
		if feature_name == 'ecfp':
		return self.contact_voxelizer._featurize_complex((lig_xyz, lig_rdk),
		(prot_xyz, prot_rdk))
		@@ -352,7 +352,6 @@ class RdkitGridFeaturizer(ComplexFeaturizer):
		return None

		time1 = time.time()
		# TODO(rbharath): Wait a minute, if we're subtract different centroids for each, have we ruined the alignment between protein and ligand. Figure this out before merging in.
		centroid = compute_centroid(ligand_xyz)
		ligand_xyz = subtract_centroid(ligand_xyz, centroid)
		protein_xyz = subtract_centroid(protein_xyz, centroid)

deepchem/feat/splif_fingerprint.py

+11 −0

Original line number	Diff line number	Diff line
		@@ -79,6 +79,13 @@ class SplifFingerprint(ComplexFeaturizer):
		each ligand atom, it identifies close protein atoms. These
		atom pairs are expanded to 2D circular fragments and a
		fingerprint for the union is turned on in the bit vector.

		This is conceptually pretty similar to
		`ContactCircularFingerprint` but computes ECFP fragments only
		for direct contacts instead of the entire contact region.

		For a macromolecular complex, returns a vector of shape
		`(2*size,)`
		"""

		def __init__(self,
		@@ -145,6 +152,10 @@ class SplifVoxelizer(ComplexFeaturizer):
		space, by assigning features to the voxel in which they
		originated. This technique may be useful for downstream
		learning methods such as convolutional networks.

		Featurizes a macromolecular complex into a tensor of shape
		`(voxels_per_edge, voxels_per_edge, voxels_per_edge, size)`
		where `voxels_per_edge = int(box_width/voxel_width)`.
		"""

		def __init__(self,

examples/qm7/qm7b_DTNN.py

+1 −1

Original line number	Diff line number	Diff line
		@@ -12,7 +12,7 @@ import tensorflow as tf
		tf.random.set_seed(123)
		import deepchem as dc

		# Load Tox21 dataset
		# Load QM7B dataset
		tasks, datasets, transformers = dc.molnet.load_qm7b_from_mat()
		train_dataset, valid_dataset, test_dataset = datasets

Admin message