def test_select_with_longer_fingerprints(pipeline_test_files):
(smiles_file, filter_output_oeb, fingerprint_output_oeb, smiles_dataset_file,
sorted_by_fingerprint_oeb) = pipeline_test_files
smiles_file.write_text("\n".join(TEST_SMILES))
dp = DancePipeline("SMILES", smiles_file)
dp.filter(_relevant_always, filter_output_oeb)
# Each fingerprint consists of (1, 1, num_atoms), to ensure that fingerprints
# are sorted correctly when the fingerprint is longer.
dp.assign_fingerprint(lambda mol: (1, 1, mol.NumAtoms()), fingerprint_output_oeb)
dp.select(3, "SMILES", smiles_dataset_file, sorted_by_fingerprint_oeb, in_memory_sorting_threshold=3)
# Check that the molecules are sorted by fingerprint.
outputted_smiles = \
[oechem.OEMolToSmiles(mol) for mol in utils.get_mols_from_oeb(sorted_by_fingerprint_oeb)]
assert outputted_smiles == \
utils.get_list_of_canonical_isomeric_smiles(["N", "N#N", "C#N", "O=C=O"]) or \
outputted_smiles == \
utils.get_list_of_canonical_isomeric_smiles(["N", "C#N", "N#N", "O=C=O"])
# Check that the correct molecules were selected.
utils.assert_smiles_in_file_are_equal(smiles_dataset_file,
utils.get_list_of_canonical_isomeric_smiles(["N", "O=C=O"]))
def _pipeline_executed_until_fingerprint(pipeline_test_files):
"""Provides a pipeline that has been executed up to and including the assign_fingerprint step.
The fingerprint function assigns a fingerprint consisting of the number of
atoms in the molecule.
Also provides several associated files.
"""
(smiles_file, filter_output_oeb, fingerprint_output_oeb, smiles_dataset_file,
sorted_by_fingerprint_oeb) = pipeline_test_files
smiles_file.write_text("\n".join(TEST_SMILES))
dp = DancePipeline("SMILES", smiles_file)
dp.filter(_relevant_always, filter_output_oeb)
dp.assign_fingerprint(lambda mol: (mol.NumAtoms(), ), fingerprint_output_oeb)
return dp, smiles_file, filter_output_oeb, fingerprint_output_oeb, smiles_dataset_file, sorted_by_fingerprint_oeb
def test_select_on_larger_dataset(pipeline_test_files):
(smiles_file, filter_output_oeb, fingerprint_output_oeb, smiles_dataset_file,
sorted_by_fingerprint_oeb) = pipeline_test_files
# The pipeline does not filter out repeated molecules, so this is okay.
smiles = ["N", "O=C=O", "C#N"] * 10
random.shuffle(smiles) # The pipeline should work for any molecule ordering.
smiles_file.write_text("\n".join(smiles))
# Pipeline execution.
dp = DancePipeline("SMILES", smiles_file)
dp.filter(_relevant_always, filter_output_oeb)
dp.assign_fingerprint(lambda mol: (mol.NumAtoms(), ), fingerprint_output_oeb)
dp.select(10, "SMILES", smiles_dataset_file, sorted_by_fingerprint_oeb, in_memory_sorting_threshold=7)
utils.assert_ordered_smiles_in_oeb_are_equal(sorted_by_fingerprint_oeb, \
utils.get_list_of_canonical_isomeric_smiles(["N"] * 10 + ["C#N"] * 10 + ["O=C=O"] * 10))
utils.assert_ordered_smiles_in_file_are_equal(smiles_dataset_file, \
utils.get_list_of_canonical_isomeric_smiles(["N", "C#N", "O=C=O"]))