def test_reordering_mols_produces_same_fprints(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
import random
rand_sdf_files = glob.glob(RAND_SDF_DIR + "/*.sdf*")
mols = list(map(mol_from_sdf, rand_sdf_files[:5]))
level = 5
fpr = fprinter.Fingerprinter(level=level,
stereo=False,
radius_multiplier=1.718,
remove_duplicate_substructs=True)
fprints1 = {}
for mol in mols:
fpr.run(conf=0, mol=mol)
fprints1[mol] = fpr.get_fingerprint_at_level(level)
random.shuffle(mols)
fpr = fprinter.Fingerprinter(level=level,
stereo=False,
radius_multiplier=1.718,
remove_duplicate_substructs=True)
fprints2 = {}
for mol in mols:
fpr.run(conf=0, mol=mol)
fprints2[mol] = fpr.get_fingerprint_at_level(level)
self.assertEqual(fprints1, fprints2)
def test_remove_dupe_substructs_makes_same_substruts_diff_shells(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
mol = mol_from_sdf(PLANAR_SDF_FILE)
level = 2
conf = mol.GetConformers()[0]
fpr = fprinter.Fingerprinter(level=level,
bits=1024,
stereo=True,
radius_multiplier=1.718,
remove_duplicate_substructs=True)
fpr.run(conf, mol)
shells_no_dupes = set(fpr.level_shells[fpr.current_level])
substructs_no_dupes = set([x.substruct for x in shells_no_dupes])
fpr = fprinter.Fingerprinter(level=level,
bits=1024,
stereo=True,
radius_multiplier=1.718,
remove_duplicate_substructs=False)
fpr.run(conf, mol)
shells_with_dupes = set(fpr.level_shells[fpr.current_level])
substructs_with_dupes = set([x.substruct for x in shells_with_dupes])
self.assertEqual(substructs_no_dupes, substructs_with_dupes)
self.assertNotEqual(shells_no_dupes, shells_with_dupes)
def test_reordering_conformers_produces_same_fprints(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
import random
rand_sdf_files = glob.glob(os.path.join(RAND_SDF_DIR, "*.sdf*"))
mol = mol_from_sdf(rand_sdf_files[0])
level = 5
fpr = fprinter.Fingerprinter(
level=level,
stereo=False,
radius_multiplier=1.718,
remove_duplicate_substructs=True,
)
conf_ids1 = [x.GetId() for x in mol.GetConformers()]
fprints1 = {}
for conf_id in conf_ids1:
fpr.run(conf_id, mol)
fprints1[conf_id] = fpr.get_fingerprint_at_level(level)
conf_ids2 = list(conf_ids1)
random.shuffle(conf_ids2)
fprints2 = {}
for conf_id in conf_ids2:
fpr.run(conf_id, mol)
fprints2[conf_id] = fpr.get_fingerprint_at_level(level)
self.assertEqual(fprints1, fprints2)
def test_quick(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
mol = mol_from_sdf(PLANAR_SDF_FILE)
level = 5
conf = mol.GetConformers()[0]
fpr = fprinter.Fingerprinter(level=level, bits=1024, stereo=True,
radius_multiplier=1.718)
fpr.run(conf, mol)
def test_initial_identifiers_assigned_correctly(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
mol = mol_from_sdf(PLANAR_SDF_FILE)
level = 0
conf = mol.GetConformers()[0]
fpr = fprinter.Fingerprinter(level=level, bits=1024, stereo=True,
radius_multiplier=1.718)
fpr.run(conf, mol)
fprint = fpr.get_fingerprint_at_level(0)
expect_ident = set([48, 124, 185, 484, 617, 674])
self.assertEqual(set(fprint.indices), expect_ident)
def test_stereoisomers_produce_equal_fingerprints_nonstereo(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
mol1 = mol_from_sdf(ENANT1_SDF_FILE)
mol2 = mol_from_sdf(ENANT2_SDF_FILE)
level = 5
fpr = fprinter.Fingerprinter(level=level, stereo=False,
radius_multiplier=1.718,
remove_duplicate_substructs=True)
fpr.run(conf=0, mol=mol1)
fp1 = fpr.get_fingerprint_at_level(level)
fpr.run(conf=0, mol=mol2)
fp2 = fpr.get_fingerprint_at_level(level)
self.assertEqual(fp1, fp2)
def test_runs_without_exception_on_random_mols(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
rand_sdf_files = glob.glob(RAND_SDF_DIR+"/*.sdf*")
rand_sdf_files = [rand_sdf_files[i] for i
in np.random.randint(len(rand_sdf_files), size=10)]
level = 5
for sdf_file in rand_sdf_files:
mol = mol_from_sdf(sdf_file)
conf = mol.GetConformers()[0]
fpr = fprinter.Fingerprinter(level=level, bits=1024, stereo=True,
radius_multiplier=1.718)
fpr.run(conf, mol)
fpr.get_fingerprint_at_level(level)
def generate_e3fp_fingerprint(sdf_file):
logging.debug(f"Generating fingerprint for file {sdf_file}")
# Load molecule from file using rdkit
molecule = Chem.SDMolSupplier(sdf_file, sanitize=False)[0]
# Calculate necessary properties
molecule.UpdatePropertyCache(strict=False)
# Set up fingerprinter and generate fingerprint for this molecule
fingerprinter = fprinter.Fingerprinter(bits=1048576)
fingerprinter.run(mol=molecule)
result = fingerprinter.get_fingerprint_at_level()
# Set name of the fingerprint to the sdf file name
result.name = sdf_file
return result
def test_repeated_runs_produce_same_results(self):
from e3fp.fingerprint import fprinter
from e3fp.conformer.util import mol_from_sdf
mol = mol_from_sdf(PLANAR_SDF_FILE)
level = 2
conf = mol.GetConformers()[0]
ref_identifiers = None
for i in range(5):
fpr = fprinter.Fingerprinter(level=level, stereo=True,
radius_multiplier=1.718)
fpr.run(conf, mol)
identifiers = sorted([x.identifier
for x in fpr.level_shells[level]])
if ref_identifiers is None:
ref_identifiers = identifiers
else:
self.assertEqual(identifiers, ref_identifiers)
def test_fingerprint_is_transform_invariant(self):
from e3fp.fingerprint import fprinter
from e3fp.fingerprint.array_ops import (
make_transform_matrix,
transform_array,
)
from e3fp.conformer.util import mol_from_sdf
mol = mol_from_sdf(PLANAR_SDF_FILE)
level = 5
conf = mol.GetConformers()[0]
ref_fp = None
atom_ids = [x.GetIdx() for x in mol.GetAtoms()]
coords = np.array(list(map(conf.GetAtomPosition, atom_ids)),
dtype=np.float)
for i in range(5):
rand_y = np.random.uniform(size=3)
rand_trans = np.random.uniform(size=3) * 100
trans_mat = make_transform_matrix(rand_trans)
rot_mat = make_transform_matrix(np.zeros(3), rand_y)
transform_mat = np.dot(trans_mat, rot_mat)
new_coords = transform_array(transform_mat, coords)
with self.assertRaises(AssertionError):
np.testing.assert_almost_equal(new_coords, coords)
for atom_id, new_coord in zip(atom_ids, new_coords):
conf.SetAtomPosition(atom_id, new_coord)
test_coords = np.array(list(map(conf.GetAtomPosition, atom_ids)),
dtype=np.float)
np.testing.assert_almost_equal(test_coords, new_coords)
fpr = fprinter.Fingerprinter(level=level,
stereo=True,
radius_multiplier=1.718)
fpr.run(conf, mol)
fp = fpr.get_fingerprint_at_level(level)
if ref_fp is None:
ref_fp = fp
else:
self.assertEqual(fp, ref_fp)
