def test_space_in_sample_name(self):
directory = path.join(path_to_module, "ngs_space_in_sample_name")
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)
file_path = path.join(directory, "SampleSheet.csv")
# Just making sure this doesn't throw an error
sample_parser.parse_sample_list(sample_sheet_file=file_path, run_data_directory=data_dir, run_data_directory_file_list=file_list)
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_valid(self):
"""
Ensure a a parsed valid directory matches the expected sample list
:return:
"""
directory = path.join(path_to_module, "fake_ngs_data")
sheet_file = path.join(directory, "SampleSheet.csv")
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 = model.Sample(
"01-1111", "", 1, {
"index": "AAAAAAAA",
"I7_Index_ID": "N01",
"sample_project": "6",
"sequencer_sample_ID": "01-1111-4004",
"I5_Index_ID": "S01",
"index2": "TTTTTTTT"
})
sequence_file_properties = {
"sequencer_sample_ID": "01-1111-4004",
"index": "AAAAAAAA",
"I7_Index_ID": "N01",
"I5_Index_ID": "S01",
"index2": "TTTTTTTT",
"description": ""
}
file_path_1 = path.join(path_to_module, "fake_ngs_data", "Alignment_1",
"some_dir", "Fastq",
"01-1111_S1_L001_R1_001.fastq.gz")
file_path_2 = path.join(path_to_module, "fake_ngs_data", "Alignment_1",
"some_dir", "Fastq",
"01-1111_S1_L001_R2_001.fastq.gz")
raw_file_list = [file_path_1, file_path_2]
res = sample_parser.parse_sample_list(
sample_sheet_file=sheet_file,
run_data_directory=data_dir,
run_data_directory_file_list=file_list)
# Check sample is the same
self.assertEqual(res[0].get_uploadable_dict(),
sample.get_uploadable_dict())
# Check sequencing file is correct
self.assertEqual(res[0].sequence_file.properties_dict,
sequence_file_properties)
self.assertEqual(res[0].sequence_file.file_list.sort(),
raw_file_list.sort())
def test_not_valid_pf_list(self):
"""
The file list in the sample sheet is invalid
:return:
"""
directory = path.join(path_to_module, "ngs_not_valid_pf_list")
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)
file_path = path.join(directory, "SampleSheet.csv")
with self.assertRaises(SequenceFileError):
res = sample_parser.parse_sample_list(sample_sheet_file=file_path, run_data_directory=data_dir, run_data_directory_file_list=file_list)
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