def test_json_read_write_success(self):
ss_file_path = os.path.join(self.temp_dir, 'ss_dis.json')
self.assertFalse(exists(ss_file_path))
write_json(self.ss_dis_dic, ss_file_path)
self.assertTrue(exists(ss_file_path))
result = read_json(ss_file_path)
self.assertEqual(self.ss_dis_dic, result)
return None
def fetch_ss_dis(dir_path):
"""Return a processed dictionary for ss_dis data.
Args:
dir_path (Unicode): The dir path where ss_dis files are located.
Returns:
ss_dis_data (dict): A dictionary of processed ss_dis data.
"""
working_path = os.path.abspath(dir_path)
ss_dis_files = _find_existing_files(working_path)
if ss_dis_files['files_to_archive']:
for name_to_archive in ss_dis_files['files_to_archive']:
path_to_archive = os.path.join(working_path, name_to_archive)
_archive_ss_data(
path_to_archive
)
if ss_dis_files['valid_raw_file']:
valid_raw_fp = os.path.join(
working_path, ss_dis_files['valid_raw_file']
)
else:
valid_raw_fp = None
if ss_dis_files['valid_json_file']:
valid_json_fp = os.path.join(
working_path, ss_dis_files['valid_json_file']
)
else:
valid_json_fp = None
# If we a valid pair exists, use the json to return a dictionary.
if valid_raw_fp and valid_json_fp:
assert os.path.isfile(valid_raw_fp)
assert os.path.isfile(valid_json_fp)
current_json_path = valid_json_fp
# Generate a companion json file if a single raw file is found.
elif valid_raw_fp:
valid_raw_fn = os.path.basename(valid_raw_fp)
assert not valid_json_fp
this_timestamp = SS_DIS_PAT.search(valid_raw_fn).group(2)
companion_json = "{}.{}.{}".format(
'ss_dis',
this_timestamp,
'json'
)
companion_json_path = os.path.join(working_path, companion_json)
ss_dict = _generate_ss_dict(valid_raw_fp)
write_json(ss_dict, companion_json_path)
current_json_path = companion_json_path
# Download new data and generate json file.
elif not (valid_raw_fp or valid_json_fp):
new_names = _new_filenames()
new_raw_path = os.path.join(working_path, new_names.raw)
new_json_path = os.path.join(working_path, new_names.json)
_download_ss_data(new_raw_path)
ss_dict = _generate_ss_dict(new_raw_path)
write_json(ss_dict, new_json_path)
current_json_path = new_json_path
elif valid_raw_fp and not valid_json_fp:
raise RuntimeError("Should not have a JSON file without a TXT file.")
else:
raise RuntimeError("Unhandled case.")
# Always return the ss_dis dictionary by reading the json
# file to ensure consistency of future runs.
ss_dis_data = read_json(current_json_path)
return ss_dis_data
def uniprot_composite(dirs):
"""Creates final UniProt DataFrame.
Create final UniProt DataFrame where the
UniProt ID provides a unique key.
Args:
dirs (ProjectFolders): A named tuple of directory paths.
"""
pdb_initial_composite_fp = os.path.join(dirs.tsv_data, "pdb_initial_composite_df.tsv")
assert os.path.isfile(pdb_initial_composite_fp)
uni_folder_path = dirs.uni_data
file_names = _create_composite_file_names()
paths = _create_composite_file_paths(uni_folder_path, file_names)
uni_composite_tsv = paths["tsv_file"]
uni_composite_yaml = paths["yaml_file"]
uni_composite_json = paths["json_file"]
if _uni_composite_file_exists(uni_folder_path):
print(
"A final uni_composite file already exists. Composite "
"function complete. (Note: remove existing uni_composite "
'files in the "{}" directory to have them '
"regenerated.".format(uni_folder_path)
)
return None
pdb_df = pd.read_csv(
pdb_initial_composite_fp, sep="\t", header=0, encoding="utf-8", keep_default_na=False, na_values=["NULL", "N/A"]
)
print("Creating the UniProt composite structure.")
uni_df = create_uni_struct(pdb_df)
print("Done creating UniProt composite structure.")
print("Validating UniProt composite structure.")
uni_pdb_validation(uni_df, pdb_df)
print("Validation complete.")
print("Assigning missing region designations.")
uni_df = create_intervals(pdb_df, uni_df)
print("Done assigning missing regions.")
assert isinstance(uni_df, pd.DataFrame)
delimiter = create_delimiter("\t")
uni_df.to_csv(uni_composite_tsv, sep=delimiter, encoding="utf-8")
uni_df.to_json(uni_composite_json, force_ascii=False)
json_data = read_json(uni_composite_json)
write_yaml(json_data, uni_composite_yaml)
print("Done writing UniProt composite files:")
print("\t{}".format(uni_composite_tsv))
print("\t{}".format(uni_composite_yaml))
print("\t{}".format(uni_composite_json))
print("This is the final UniProt ID DataFrame.")
return None
