def test_build_valid_with_description_field(self):
"""
When given a valid directory, ensure a valid SequencingRun is built with Projects, Samples, ect
:return:
"""
directory = path.join(path_to_module, "iseq_with_desc_field")
sheet_file = path.join(directory, "SampleSheet.csv")
meta_data = sample_parser.parse_metadata(sheet_file)
data_dir = path.join(directory, parsers.miniseq.Parser.get_relative_data_directory())
data_dir = data_dir.replace("*", "some_dir")
file_list = parsers.common.get_file_list(data_dir)
sample_list = sample_parser.parse_sample_list(sample_sheet_file=sheet_file,
run_data_directory=data_dir,
run_data_directory_file_list=file_list)
sequencing_run = parsers.common.build_sequencing_run_from_samples(sample_list, meta_data)
# Returns a SequencingRun
self.assertEqual(type(sequencing_run), model.SequencingRun)
# Includes a single project
self.assertEqual(len(sequencing_run.project_list), 1)
# is of type Project
self.assertEqual(type(sequencing_run.project_list[0]), model.Project)
# Project has 3 samples
self.assertEqual(len(sequencing_run.project_list[0].sample_list), 3)
# samples are of type Sample
self.assertEqual(type(sequencing_run.project_list[0].sample_list[0]), model.Sample)
# samples have correct description
self.assertEqual(sequencing_run.project_list[0].sample_list[0].description, "desc1")
# samples have SequenceFile
self.assertEqual(type(sequencing_run.project_list[0].sample_list[0].sequence_file), model.SequenceFile)
def test_parse_metadata_paired_valid(self, mock_csv_reader):
"""
When given a valid directory, ensure valid metadata is built
paired end reads
:return:
"""
h_field_values = (
"Local Run Manager Analysis Id,4004\n"
"Experiment Name,Some_Test_Data\n"
"Date,2015-05-14\n"
"Workflow,GenerateFastQWorkflow\n"
"Description,12-34\n"
"Chemistry,Yes\n"
)
reads = (
"151\n"
"151\n"
)
d_headers = ("Sample_ID,Sample_Name,"
"I7_Index_ID,index,I5_Index_ID,index2,Sample_Project")
d_field_values = ("15-0318-4004,15-0318,N701,TAAGGCGA,S502,CTCTCTAT,203\n"
"15-0455-4004,15-0455,N701,TAAGGCGA,S503,TATCCTCT,203\n"
"15-0462-4004,15-0462,N701,TAAGGCGA,S505,GTAAGGAG,203\n")
file_contents_str = (
"[Header]\n"
"{h_field_values}\n"
"[Reads]\n"
"{reads}\n"
"[Data]\n"
"{d_headers}\n"
"{d_field_values}"
).format(h_field_values=h_field_values,
reads=reads,
d_headers=d_headers,
d_field_values=d_field_values)
# converts string as a pseudo file / memory file
sample_sheet_file = StringIO(file_contents_str)
# the call to get_csv_reader() inside parse_samples() will return
# items inside side_effect
mock_csv_reader.side_effect = [reader(sample_sheet_file)]
metadata = sample_parser.parse_metadata(None)
# The meta data we care about the most
self.assertEqual(metadata['readLengths'], "151")
self.assertEqual(metadata['layoutType'], "PAIRED_END")
# Other meta data should also be here
self.assertEqual(metadata['localrunmanager'], "4004")
self.assertEqual(metadata['experimentName'], "Some_Test_Data")
self.assertEqual(metadata['date'], "2015-05-14")
self.assertEqual(metadata['workflow'], "GenerateFastQWorkflow")
self.assertEqual(metadata['description'], "12-34")
self.assertEqual(metadata['chemistry'], "Yes")
def test_parse_metadata(self):
"""
Testing the parsing meta data with actual files, instead of mocked files
:return:
"""
sheet_file = path.join(path_to_module, "fake_ngs_data",
"SampleSheet.csv")
meta_data = sample_parser.parse_metadata(sheet_file)
correct_metadata = {"readLengths": "151",
"workflow": "GenerateFastQWorkflow",
"localrunmanager": "4004",
"date": "10/15/2013",
"chemistry": "Amplicon",
"description": "Superbug",
"experimentName": '1',
"layoutType": "PAIRED_END"}
self.assertEqual(correct_metadata, meta_data)
def get_sequencing_run(sample_sheet,
run_data_directory=None,
run_data_directory_file_list=None):
"""
Does local validation on the integrity of the run directory / sample sheet
Throws a ValidationError with a validation result attached if it cannot make a sequencing run
:param sample_sheet: Sample Sheet File
:param run_data_directory: Optional: Directory (including run directory) to data files.
Can be provided for bypassing os calls when developing on cloud systems
:param run_data_directory_file_list: Optional: List of files in data directory.
Can be provided for bypassing os calls when developing on cloud systems
:return: SequencingRun
"""
# get data directory and file list
validation_result = model.ValidationResult()
try:
if run_data_directory is None:
run_data_directory = Parser.get_full_data_directory(
sample_sheet)
if run_data_directory_file_list is None:
run_data_directory_file_list = common.get_file_list(
run_data_directory)
except exceptions.DirectoryError as error:
validation_result.add_error(error)
logging.error("Errors occurred while parsing files")
raise exceptions.ValidationError(
"Errors occurred while parsing files", validation_result)
# Try to get the sample sheet, validate that the sample sheet is valid
validation_result = validation.validate_sample_sheet(sample_sheet)
if not validation_result.is_valid():
logging.error("Errors occurred while getting sample sheet")
raise exceptions.ValidationError(
"Errors occurred while getting sample sheet",
validation_result)
# Try to parse the meta data from the sample sheet, throw validation error if errors occur
validation_result = model.ValidationResult()
try:
run_metadata = sample_parser.parse_metadata(sample_sheet)
except exceptions.SampleSheetError as error:
validation_result.add_error(error)
logging.error("Errors occurred while parsing metadata")
raise exceptions.ValidationError(
"Errors occurred while parsing metadata", validation_result)
# Try to build sequencing run from sample sheet & meta data, raise validation error if errors occur
try:
sample_list = sample_parser.parse_sample_list(
sample_sheet, run_data_directory, run_data_directory_file_list)
sequencing_run = common.build_sequencing_run_from_samples(
sample_list, run_metadata)
except exceptions.SequenceFileError as error:
validation_result.add_error(error)
logging.error(
"Errors occurred while building sequence run from sample sheet"
)
raise exceptions.ValidationError(
"Errors occurred while building sequence run from sample sheet",
validation_result)
return sequencing_run
