def test_append_fps(self):
in_file = os.path.join(TESTS_DIR, "data/10mols.h5")
fpe = FPSim2Engine(in_file,
in_memory_fps=False,
storage_backend="pytables")
fpe.storage.append_fps([["CC", 11], ["CCC", 12], ["CCCC", 13]])
fpe = FPSim2Engine(in_file, in_memory_fps=True)
assert fpe.fps.shape[0] == 13
def test_n_delete_fps(self):
in_file = os.path.join(TESTS_DIR, "data/10mols.h5")
fpe = FPSim2Engine(in_file,
in_memory_fps=False,
storage_backend="pytables")
fpe.storage.delete_fps([11, 12, 13])
sort_db_file(in_file)
fpe = FPSim2Engine(in_file)
assert fpe.fps.shape[0] == 10
assert fpe.fps[0][-1] == 35
assert fpe.fps[-1][-1] == 48
def get_fpsim_engine():
global FPSIM_ENGINE, FPSIM2_FILE_PATH
if FPSIM_ENGINE is None:
t_ini = time.time()
FPSIM_ENGINE = FPSim2Engine(FPSIM2_FILE_PATH)
print('FPSIM2 FILE LOADED IN {0} SECS'.format(time.time() - t_ini))
return FPSIM_ENGINE
def test_calc_popcnt_bins():
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file,
in_memory_fps=True,
fps_sort=True,
storage_backend="pytables")
assert fpe.popcnt_bins == popcnt_bins
def test_on_disk_substructure(n_workers):
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file,
in_memory_fps=False,
storage_backend="pytables")
results = fpe.on_disk_substructure(query_smi, n_workers=n_workers)
res = np.array(np.array([1], dtype="<u4"))
np.testing.assert_array_equal(results, res)
with pytest.raises(Exception):
fpe.fps
def get_similar_mols(kekule_smiles: list, model: str):
sim_vals = []
fp_dict_path = ''.join(['./train_data/', model, '.h5'])
fp_dict_path = path.abspath(path.join(os.getcwd(), fp_dict_path))
fp_engine = FPSim2Engine(fp_dict_path)
for smi in kekule_smiles:
res = fp_engine.on_disk_similarity(smi, 0.01)
sim_vals.append(res[0][1])
return sim_vals
def test_create_db_file_sdf():
in_file = os.path.join(TESTS_DIR, "data/10mols.sdf")
out_file = os.path.join(TESTS_DIR, "data/10mols_sdf.h5")
create_db_file(in_file, out_file, FP_TYPE, FP_PARAMS, mol_id_prop="mol_id")
fpe = FPSim2Engine(out_file, storage_backend="pytables")
fp_type, fp_params, _ = fpe.storage.read_parameters()
assert fp_type == FP_TYPE
assert fp_params["radius"] == FP_PARAMS["radius"]
assert fp_params["nBits"] == FP_PARAMS["nBits"]
assert fpe.fps.shape[0] == 10
def test_create_db_file_list():
out_file = os.path.join(TESTS_DIR, "data/10mols_list.h5")
create_db_file([["CC", 1], ["CCC", 2], ["CCCC", 3]], out_file, FP_TYPE,
FP_PARAMS)
fpe = FPSim2Engine(out_file, storage_backend="pytables")
fp_type, fp_params, _ = fpe.storage.read_parameters()
assert fp_type == FP_TYPE
assert fp_params["radius"] == FP_PARAMS["radius"]
assert fp_params["nBits"] == FP_PARAMS["nBits"]
assert fpe.fps.shape[0] == 3
def test_tversky(n_workers):
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file, storage_backend="pytables")
results = fpe.tversky(query_smi, 0.85, 0.5, 0.5, n_workers=n_workers)
res = np.array(
[(1, 1.0), (6, 0.85057473)],
dtype={
"names": ["mol_id", "coeff"],
"formats": ["<u4", "<f4"]
},
)
np.testing.assert_array_equal(results, res)
def test_validate_against_rdkit():
with open('test/10mols.smi') as f:
smiles = f.readlines()
fps = [Chem.rdMolDescriptors.GetMorganFingerprintAsBitVect(Chem.MolFromSmiles(smi),
radius=2, nBits=2048) for smi in smiles]
query = Chem.rdMolDescriptors.GetMorganFingerprintAsBitVect(Chem.MolFromSmiles(query_smi),
radius=2, nBits=2048)
rdresults = sorted([DataStructs.TanimotoSimilarity(query, fp) for fp in fps], reverse=True)
fpe = FPSim2Engine('test/10mols.h5')
results = fpe.similarity(query_smi, 0.0, n_workers=1)['coeff']
for rds, fpss in zip(rdresults, results):
assert True == math.isclose(rds, fpss, rel_tol=1e-7)
def test_on_disk_similarity(n_workers):
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file,
in_memory_fps=False,
storage_backend="pytables")
with pytest.raises(Exception):
fpe.fps
results = fpe.on_disk_similarity(query_smi, 0.7, n_workers=n_workers)
r = np.array(
[(1, 1.0), (6, 0.74), (7, 0.735849), (5, 0.72549)],
dtype={
"names": ["mol_id", "coeff"],
"formats": ["<u4", "<f4"],
"offsets": [4, 8],
"itemsize": 12,
},
)
np.testing.assert_array_almost_equal(results["coeff"], r["coeff"])
def test_validate_against_rdkit():
in_file_smi = os.path.join(TESTS_DIR, "data/10mols.smi")
in_file_h5 = os.path.join(TESTS_DIR, "data/test.h5")
with open(in_file_smi) as f:
smiles = f.readlines()
fps = [
Chem.rdMolDescriptors.GetMorganFingerprintAsBitVect(
Chem.MolFromSmiles(smi), **FP_PARAMS) for smi in smiles
]
query = Chem.rdMolDescriptors.GetMorganFingerprintAsBitVect(
Chem.MolFromSmiles(query_smi), **FP_PARAMS)
rdresults = sorted(
[DataStructs.TanimotoSimilarity(query, fp) for fp in fps],
reverse=True)
fpe = FPSim2Engine(in_file_h5, storage_backend="pytables")
results = fpe.similarity(query_smi, 0.0, n_workers=1)["coeff"]
for rds, fpss in zip(rdresults, results):
assert math.isclose(rds, fpss, rel_tol=1e-7)
def test_symmetric_matrix(n_workers):
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file, storage_backend="pytables")
csr_matrix = fpe.symmetric_distance_matrix(0.0, n_workers=n_workers)
np.testing.assert_array_equal(MATRIX, csr_matrix.todense())
def test_substructure(n_workers):
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file, storage_backend="pytables")
results = fpe.substructure(query_smi, n_workers=n_workers)
res = np.array(np.array([1], dtype="<u4"))
np.testing.assert_array_equal(results, res)
def test_similarity(n_workers):
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file, storage_backend="pytables")
results = fpe.similarity(query_smi, 0.7, n_workers=n_workers)
assert results.shape[0] == 4
assert list(results[0]) == [1, 1.0]
def test_sort_db_file(self):
in_file = os.path.join(TESTS_DIR, "data/10mols.h5")
sort_db_file(in_file)
fpe = FPSim2Engine(in_file, storage_backend="pytables")
assert fpe.fps[0][-1] == 2
assert fpe.fps[-1][-1] == 48
def test_on_disk_search():
fpe = FPSim2Engine('test/10mols.h5', in_memory_fps=False)
results = fpe.on_disk_similarity(query_smi, 0.7, chunk_size=100000, n_workers=2)
assert results.shape[0] == 4
assert list(results[0]) == [1, 1.0]
def test_search():
fpe = FPSim2Engine('test/10mols.h5')
results = fpe.similarity(query_smi, 0.7, n_workers=1)
assert results.shape[0] == 4
assert list(results[0]) == [1, 1.0]
def test_load_fps():
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file, storage_backend="pytables")
assert fpe.fps.shape[0] == 10
assert fpe.fps.shape[1] == 34
assert fpe.popcnt_bins != []
import numpy as np
from FPSim2.FPSim2lib import (
PyPopcount,
SortResults,
SimilaritySearch,
SubstructureScreenout,
)
from FPSim2 import FPSim2Engine
import os
TESTS_DIR = os.path.dirname(os.path.abspath(__file__))
fpe = FPSim2Engine(os.path.join(TESTS_DIR, "data/test.h5"))
def test_PyPopcount():
assert PyPopcount(fpe.fps[0]) == 39
assert PyPopcount(fpe.fps[1]) == 45
assert PyPopcount(fpe.fps[2]) == 43
def test_tanimoto_SimilaritySearch():
res = SimilaritySearch(fpe.fps[0], fpe.fps, 0.0, 0.0, 0.0, 0, 0,
fpe.fps.shape[0])
np.testing.assert_array_almost_equal(
res[0:5]["coeff"],
np.array(
[
(0, 4, 1.0),
(3, 6, 0.63829786),
(5, 5, 0.625),
(7, 2, 0.5882353),
def test_load_fps_sort():
in_file = os.path.join(TESTS_DIR, "data/test.h5")
fpe = FPSim2Engine(in_file, storage_backend="pytables")
fpe2 = FPSim2Engine(in_file, fps_sort=True)
assert fpe.popcnt_bins == fpe2.popcnt_bins