def getChannels(mol, aromaticNitrogen=False, version=2, validitychecks=True):
from moleculekit.smallmol.smallmol import SmallMol
from moleculekit.molecule import Molecule
mol = mol.copy()
if isinstance(mol, SmallMol):
channels = _getPropertiesRDkit(mol)
elif isinstance(mol, Molecule):
if version == 1:
channels = _getAtomtypePropertiesPDBQT(mol)
elif version == 2:
from moleculekit.tools.atomtyper import (
getFeatures,
getPDBQTAtomTypesAndCharges,
)
mol.atomtype, mol.charge = getPDBQTAtomTypesAndCharges(
mol,
aromaticNitrogen=aromaticNitrogen,
validitychecks=validitychecks)
channels = getFeatures(mol)
if channels.dtype == bool:
# Calculate for each channel the atom sigmas
sigmas = _getChannelRadii(mol.get("element"))
channels = sigmas[:, np.newaxis] * channels.astype(float)
return channels, mol
def getChannels(mol,
aromaticNitrogen=False,
version=2,
validitychecks=True,
protein=True):
#from moleculekit.smallmol.smallmol import SmallMol
mol = mol.copy()
if isinstance(mol, SmallMol):
channels = _getPropertiesRDkit(mol)
elif isinstance(mol, Molecule):
if version == 1:
channels = _getAtomtypePropertiesPDBQT(mol)
elif version == 2:
mol.atomtype, mol.charge = getPDBQTAtomTypesAndCharges(
mol,
aromaticNitrogen=aromaticNitrogen,
validitychecks=validitychecks)
residues = mol.resname # residue name for each atom
amacid_features = np.stack(
[amino_acid_feature(r) for r in residues], axis=0)
elements = mol.element
els = np.stack([get_atomic_feature(e) for e in elements], axis=0)
from tempfile import NamedTemporaryFile
tmpmol2 = NamedTemporaryFile(suffix='.pdb').name
mol.write(tmpmol2)
s_atoms = process_arpeggio(tmpmol2)
os.remove(tmpmol2)
channels = getFeatures(mol)
# we take only
channels = channels[:, features_to_keep]
#assert len(channels) == len(s_atoms)
# the order of s_atoms matches the order of channels
arpeggio_features = []
for i, atom in enumerate(s_atoms):
raw_features = sorted(tuple(s_atoms[i].atom_types))
feature_vector = np.zeros(len(arpeggio_atomtypes_unique),
dtype=bool)
for rf in raw_features:
#print(rf)
if rf in arpeggio_atomtypes_unique:
feature_vector[arpeggio_atomtypes_unique[rf]] = 1
arpeggio_features.append(feature_vector)
if feature_vector[-1] != 0:
print(raw_features, feature_vector)
arpeggio_features = np.stack(arpeggio_features, axis=0)
#print('aaa', channels.shape, arpeggio_features.shape,amacid_features.shape,els.shape)
protein_feature = np.ones(
shape=(len(channels), 1),
dtype=bool) # 1 for protein, 0 for ligand
channels = np.concatenate([
channels, amacid_features, arpeggio_features, els,
protein_feature
],
axis=-1)
#channels = np.concatenate([arpeggio_features,amacid_features,els], axis=-1 )
'''
# check that the order of atoms is the same
coords = mol.get('coords')
serials = mol.get('serial')
for i, elem in enumerate(mol.get('element')):
#print(elem, s_atoms[i].get_name())
print(sorted(tuple(s_atoms[i].atom_types)))
try:
assert np.allclose(coords[i],s_atoms[i].get_coord(), atol=1e-3)
except:
print(coords[i],s_atoms[i].get_coord())
assert serials[i] == s_atoms[i].get_serial_number()
print('11111111111', len(channels), len(s_atoms))
'''
if channels.dtype == bool:
# Calculate for each channel the atom sigmas
sigmas = _getChannelRadii(mol.get('element'))
channels = sigmas[:, np.newaxis] * channels.astype(float)
#print(channels.shape,arpeggio_features.shape)
#assert channels.shape[0] == arpeggio_features.shape[0]
#channels = np.concatenate([channels, arpeggio_features], axis=-1 )
#print('bbb', channels.shape, (1000*np.mean(channels, axis=0)).astype(int))
#print('ccc', (1000*np.mean(channels[:,:12], axis=0)).astype(int))
return channels, mol