def test_setup_analysis_directory_structure(self, mock_id, mock_parse,
mock_path, mock_makedir):
fc_dir = '/ngi2016003/201103_A00187_0332_AHFCFLDSXX'
mock_parse.return_value = {
'fc_dir':
fc_dir,
'fc_full_id':
'201103_A00187_0332_AHFCFLDSXX',
'projects': [{
'project_name': 'S.One_20_01',
'project_original_name': 'something',
'samples': [{
'sample_name': 'one'
}]
}]
}
mock_id.return_value = 'P12345'
projects_to_analyze = {}
expected_project = 'S.One_20_01'
got_projects = setup_analysis_directory_structure(fc_dir,
projects_to_analyze,
create_files=False)
got_project = got_projects[
'/lupus/ngi/staging/wildwest/ngi2016001/nobackup/NGI/DATA/P12345']
self.assertEqual(expected_project, got_project.name)
def main(demux_fcid_dirs, restrict_to_projects=None, restrict_to_samples=None,
force_update=False, workflow="NGI", already_parsed=False,
config=None, config_file_path=None):
if force_update: force_update = validate_force_update()
if not restrict_to_projects: restrict_to_projects = []
if not restrict_to_samples: restrict_to_samples = []
demux_fcid_dirs_set = set(demux_fcid_dirs)
# Sort/copy each raw demux FC into project/sample/fcid format -- "analysis-ready"
projects_to_analyze = dict()
if already_parsed: # Starting from Project/Sample/Libprep/Seqrun tree format
for demux_fcid_dir in demux_fcid_dirs_set:
p = recreate_project_from_filesystem(demux_fcid_dir)
projects_to_analyze[p.name] = p
else: # Raw illumina flowcell
for demux_fcid_dir in demux_fcid_dirs_set:
# These will be a bunch of Project objects each containing Samples, FCIDs, lists of fastq files
projects_to_analyze = setup_analysis_directory_structure(demux_fcid_dir,
projects_to_analyze,
restrict_to_projects,
restrict_to_samples,
create_files=False,
config=config)
if not projects_to_analyze:
sys.exit("Quitting: no projects found to process in flowcells {}"
"or there was an error gathering required "
"information.".format(",".join(demux_fcid_dirs_set)))
else:
# Don't need the dict functionality anymore; revert to list
projects_to_analyze = projects_to_analyze.values()
for project in projects_to_analyze:
try:
create_charon_entries_from_project(project, workflow=workflow, force_overwrite=force_update)
except Exception as e:
print(e, file=sys.stderr)
def main(demux_fcid_dirs,
restrict_to_projects=None,
restrict_to_samples=None,
best_practice_analysis=None,
sequencing_facility=None,
already_parsed=False,
force_update=False,
delete_existing=False,
force_create_project=False,
config=None,
config_file_path=None):
if force_update: force_update = validate_force_update()
if delete_existing: delete_existing = validate_delete_existing()
if not restrict_to_projects: restrict_to_projects = []
if not restrict_to_samples: restrict_to_samples = []
demux_fcid_dirs_set = set(demux_fcid_dirs)
projects_to_analyze = dict()
if already_parsed: # Starting from Project/Sample/Libprep/Seqrun tree format
for demux_fcid_dir in demux_fcid_dirs_set:
p = recreate_project_from_filesystem(
demux_fcid_dir, force_create_project=force_create_project)
projects_to_analyze[p.name] = p
else: # Raw illumina flowcell
for demux_fcid_dir in demux_fcid_dirs_set:
projects_to_analyze = setup_analysis_directory_structure(
demux_fcid_dir,
projects_to_analyze,
restrict_to_projects,
restrict_to_samples,
config=config)
if not projects_to_analyze:
sys.exit("Quitting: no projects found to process in flowcells {}"
"or there was an error gathering required "
"information.".format(",".join(demux_fcid_dirs_set)))
else:
projects_to_analyze = projects_to_analyze.values()
for project in projects_to_analyze:
try:
create_charon_entries_from_project(
project,
best_practice_analysis=best_practice_analysis,
sequencing_facility=sequencing_facility,
force_overwrite=force_update,
delete_existing=delete_existing)
except Exception as e:
print(e, file=sys.stderr)
def main(demux_fcid_dirs, restrict_to_projects=None, restrict_to_samples=None,
best_practice_analysis=None, sequencing_facility=None,
already_parsed=False,
force_update=False, delete_existing=False,
force_create_project=False,
config=None, config_file_path=None):
if force_update: force_update = validate_force_update()
if delete_existing: delete_existing = validate_delete_existing()
if not restrict_to_projects: restrict_to_projects = []
if not restrict_to_samples: restrict_to_samples = []
demux_fcid_dirs_set = set(demux_fcid_dirs)
projects_to_analyze = dict()
if already_parsed: # Starting from Project/Sample/Libprep/Seqrun tree format
for demux_fcid_dir in demux_fcid_dirs_set:
p = recreate_project_from_filesystem(demux_fcid_dir,
force_create_project=force_create_project)
projects_to_analyze[p.name] = p
else: # Raw illumina flowcell
for demux_fcid_dir in demux_fcid_dirs_set:
projects_to_analyze = setup_analysis_directory_structure(demux_fcid_dir,
projects_to_analyze,
restrict_to_projects,
restrict_to_samples,
config=config)
if not projects_to_analyze:
sys.exit("Quitting: no projects found to process in flowcells {}"
"or there was an error gathering required "
"information.".format(",".join(demux_fcid_dirs_set)))
else:
projects_to_analyze = projects_to_analyze.values()
for project in projects_to_analyze:
try:
create_charon_entries_from_project(project, best_practice_analysis=best_practice_analysis,
sequencing_facility=sequencing_facility,
force_overwrite=force_update,
delete_existing=delete_existing)
except Exception as e:
print(e, file=sys.stderr)
if args.delete_existing: args.delete_existing = validate_delete_existing()
if not args.restrict_to_projects: args.restrict_to_projects = []
if not args.restrict_to_samples: args.restrict_to_samples = []
organize_fc_dirs_set = set(args.organize_fc_dirs)
projects_to_analyze = dict()
## NOTE this bit of code not currently in use but could use later
#if args.already_parsed: # Starting from Project/Sample/Libprep/Seqrun tree format
# for organize_fc_dir in organize_fc_dirs_set:
# p = recreate_project_from_filesystem(organize_fc_dir,
# force_create_project=args.force_create_project)
# projects_to_analyze[p.name] = p
#else: # Raw illumina flowcell
for organize_fc_dir in organize_fc_dirs_set:
projects_to_analyze = setup_analysis_directory_structure(fc_dir=organize_fc_dir,
projects_to_analyze=projects_to_analyze,
restrict_to_projects=args.restrict_to_projects,
restrict_to_samples=args.restrict_to_samples,
fallback_libprep=args.fallback_libprep,
quiet=args.quiet)
if not projects_to_analyze:
raise ValueError('No projects found to process in flowcells '
'"{}" or there was an error gathering required '
'information.'.format(",".join(organize_fc_dirs_set)))
else:
projects_to_analyze = projects_to_analyze.values()
for project in projects_to_analyze:
try:
create_charon_entries_from_project(project,
best_practice_analysis=args.best_practice_analysis,
sequencing_facility=args.sequencing_facility,
force_overwrite=args.force_update,
delete_existing=args.delete_existing)
