def main(input_filename, map_filename, output_filename, column_indices=None):
"""
Get regression targets.
Parameters
----------
input_filename : str
PCBA data filename.
map_filename : str
ID->SMILES map filename.
output_filename : str
Output filename.
column_indices : list, optional
Column indices to include. If None, compounds are classified by
activity.
"""
parser = PcbaParser(input_filename, map_filename,
column_indices=column_indices)
if column_indices is not None:
print "Extracting data from the following columns:"
for col in parser.get_column_names():
print '\t', col
smiles, targets = parser.get_targets()
# print the fraction of valid assay records that were found in the map
total = np.count_nonzero(~np.isnan(parser.read_data().PUBCHEM_CID))
print '{}/{} records matched'.format(len(targets), total)
# save SMILES and targets
write_pickle({'smiles': smiles, 'targets': targets}, output_filename)
class TestPcbaParser(unittest.TestCase):
"""
Tests for PcbaParser.
"""
def setUp(self):
"""
Set up tests.
"""
self.temp_dir = tempfile.mkdtemp()
self.map = {
'CID645443':
'Cc1ccc(-n2c3c(cc(C(=O)Nc4cccc(C)n4)c2=O)C(=O)CCC3)cc1',
'CID2997889': 'CC(C)(C)C(=O)Nc1ccc(-c2cn3ccsc3n2)cc1',
'CID2244': 'CC(=O)Oc1ccccc1C(=O)O',
'CID2662': 'Cc1ccc(-c2cc(C(F)(F)F)nn2-c2ccc(S(N)(=O)=O)cc2)cc1',
'CID3672': 'CC(C)Cc1ccc(C(C)C(=O)O)cc1'
}
_, self.map_filename = tempfile.mkstemp(dir=self.temp_dir,
suffix='.pkl')
write_pickle(self.map, self.map_filename)
# use a subset of AID588342
# note that CID 654924 is duplicated
this_dir = os.path.split(os.path.realpath(__file__))[0]
self.data_filename = os.path.join(this_dir, 'data/test_pcba_data.csv')
# set up parser
self.engine = PcbaParser(self.data_filename, self.map_filename)
def tearDown(self):
"""
Clean up tests.
"""
shutil.rmtree(self.temp_dir)
def test_get_targets_classification(self):
"""
Test PcbaParser.get_targets with classification.
"""
smiles, targets = self.engine.get_targets()
assert len(smiles) == len(targets) == 2
idx = np.where(smiles == self.map['CID2997889'])[0][0]
assert targets[idx] # marked Active
idx = np.where(smiles == self.map['CID645443'])[0][0]
assert not targets[idx] # marked Inactive
def test_get_targets_regression(self):
"""
Test PcbaParser.get_targets with regression.
"""
columns = [7, 8, 12, 14, 15, 20, 22, 23, 24, 25, 26]
self.engine = PcbaParser(self.data_filename,
self.map_filename,
column_indices=columns)
smiles, targets = self.engine.get_targets()
idx = np.where(smiles == self.map['CID2997889'])[0][0]
assert not np.any(np.isnan(targets[idx]))
idx = np.where(smiles == self.map['CID645443'])[0][0]
assert np.any(np.isnan(targets[idx])) # will have NaNs
class TestPcbaParser(unittest.TestCase):
"""
Tests for PcbaParser.
"""
def setUp(self):
"""
Set up tests.
"""
self.temp_dir = tempfile.mkdtemp()
self.map = {
"CID645443": "Cc1ccc(-n2c3c(cc(C(=O)Nc4cccc(C)n4)c2=O)C(=O)CCC3)cc1",
"CID2997889": "CC(C)(C)C(=O)Nc1ccc(-c2cn3ccsc3n2)cc1",
"CID2244": "CC(=O)Oc1ccccc1C(=O)O",
"CID2662": "Cc1ccc(-c2cc(C(F)(F)F)nn2-c2ccc(S(N)(=O)=O)cc2)cc1",
"CID3672": "CC(C)Cc1ccc(C(C)C(=O)O)cc1",
}
_, self.map_filename = tempfile.mkstemp(dir=self.temp_dir, suffix=".pkl")
write_pickle(self.map, self.map_filename)
# use a subset of AID588342
# note that CID 654924 is duplicated
this_dir = os.path.split(os.path.realpath(__file__))[0]
self.data_filename = os.path.join(this_dir, "data/test_pcba_data.csv")
# set up parser
self.engine = PcbaParser(self.data_filename, self.map_filename)
def tearDown(self):
"""
Clean up tests.
"""
shutil.rmtree(self.temp_dir)
def test_get_targets_classification(self):
"""
Test PcbaParser.get_targets with classification.
"""
smiles, targets = self.engine.get_targets()
assert len(smiles) == len(targets) == 2
idx = np.where(smiles == self.map["CID2997889"])[0][0]
assert targets[idx] # marked Active
idx = np.where(smiles == self.map["CID645443"])[0][0]
assert not targets[idx] # marked Inactive
def test_get_targets_regression(self):
"""
Test PcbaParser.get_targets with regression.
"""
columns = [7, 8, 12, 14, 15, 20, 22, 23, 24, 25, 26]
self.engine = PcbaParser(self.data_filename, self.map_filename, column_indices=columns)
smiles, targets = self.engine.get_targets()
idx = np.where(smiles == self.map["CID2997889"])[0][0]
assert not np.any(np.isnan(targets[idx]))
idx = np.where(smiles == self.map["CID645443"])[0][0]
assert np.any(np.isnan(targets[idx])) # will have NaNs
def test_read_data(self):
"""
Test Nci60Parser.read_data.
"""
df = self.engine.read_data()
fixed_count = df.count().values.sum() # count excluding NaNs
# use PcbaParser to read data (w/o proper NaN handling)
engine = PcbaParser(self.data_filename, self.map_filename, delimiter="\t", primary_key="NSC", id_prefix="NSC")
df = engine.read_data()
broken_count = df.count().values.sum()
assert fixed_count < broken_count
def test_get_targets_regression(self):
"""
Test PcbaParser.get_targets with regression.
"""
columns = [7, 8, 12, 14, 15, 20, 22, 23, 24, 25, 26]
self.engine = PcbaParser(self.data_filename,
self.map_filename,
column_indices=columns)
smiles, targets = self.engine.get_targets()
idx = np.where(smiles == self.map['CID2997889'])[0][0]
assert not np.any(np.isnan(targets[idx]))
idx = np.where(smiles == self.map['CID645443'])[0][0]
assert np.any(np.isnan(targets[idx])) # will have NaNs
def test_read_data(self):
"""
Test Nci60Parser.read_data.
"""
df = self.engine.read_data()
fixed_count = df.count().values.sum() # count excluding NaNs
# use PcbaParser to read data (w/o proper NaN handling)
engine = PcbaParser(self.data_filename,
self.map_filename,
delimiter='\t',
primary_key='NSC',
id_prefix='NSC')
df = engine.read_data()
broken_count = df.count().values.sum()
assert fixed_count < broken_count
def test_get_targets_regression(self):
"""
Test PcbaParser.get_targets with regression.
"""
columns = [7, 8, 12, 14, 15, 20, 22, 23, 24, 25, 26]
self.engine = PcbaParser(self.data_filename, self.map_filename, column_indices=columns)
smiles, targets = self.engine.get_targets()
idx = np.where(smiles == self.map["CID2997889"])[0][0]
assert not np.any(np.isnan(targets[idx]))
idx = np.where(smiles == self.map["CID645443"])[0][0]
assert np.any(np.isnan(targets[idx])) # will have NaNs
def setUp(self):
"""
Set up tests.
"""
self.temp_dir = tempfile.mkdtemp()
self.map = {
'CID645443':
'Cc1ccc(-n2c3c(cc(C(=O)Nc4cccc(C)n4)c2=O)C(=O)CCC3)cc1',
'CID2997889': 'CC(C)(C)C(=O)Nc1ccc(-c2cn3ccsc3n2)cc1',
'CID2244': 'CC(=O)Oc1ccccc1C(=O)O',
'CID2662': 'Cc1ccc(-c2cc(C(F)(F)F)nn2-c2ccc(S(N)(=O)=O)cc2)cc1',
'CID3672': 'CC(C)Cc1ccc(C(C)C(=O)O)cc1'
}
_, self.map_filename = tempfile.mkstemp(dir=self.temp_dir,
suffix='.pkl')
write_pickle(self.map, self.map_filename)
# use a subset of AID588342
# note that CID 654924 is duplicated
this_dir = os.path.split(os.path.realpath(__file__))[0]
self.data_filename = os.path.join(this_dir, 'data/test_pcba_data.csv')
# set up parser
self.engine = PcbaParser(self.data_filename, self.map_filename)
def setUp(self):
"""
Set up tests.
"""
self.temp_dir = tempfile.mkdtemp()
self.map = {
"CID645443": "Cc1ccc(-n2c3c(cc(C(=O)Nc4cccc(C)n4)c2=O)C(=O)CCC3)cc1",
"CID2997889": "CC(C)(C)C(=O)Nc1ccc(-c2cn3ccsc3n2)cc1",
"CID2244": "CC(=O)Oc1ccccc1C(=O)O",
"CID2662": "Cc1ccc(-c2cc(C(F)(F)F)nn2-c2ccc(S(N)(=O)=O)cc2)cc1",
"CID3672": "CC(C)Cc1ccc(C(C)C(=O)O)cc1",
}
_, self.map_filename = tempfile.mkstemp(dir=self.temp_dir, suffix=".pkl")
write_pickle(self.map, self.map_filename)
# use a subset of AID588342
# note that CID 654924 is duplicated
this_dir = os.path.split(os.path.realpath(__file__))[0]
self.data_filename = os.path.join(this_dir, "data/test_pcba_data.csv")
# set up parser
self.engine = PcbaParser(self.data_filename, self.map_filename)
