def main(inbox=None, num_days=14, genotype_files=None, config=None, config_file_path=None):
if genotype_files:
gt_files_valid = [os.path.abspath(gt_file) for gt_file in genotype_files]
else:
if not inbox:
try:
inboxes = config["environment"]["flowcell_inbox"]
except (KeyError, TypeError):
raise ValueError("No path to delivery inbox specified by argument "
"or in configuration file ({}). Exiting.".format(config_file_path))
for inbox in inboxes:
inbox = os.path.abspath(inbox)
# Convert to seconds
cutoff_age = time.time() - (int(num_days) * 24 * 60 * 60)
LOG.info("Searching for genotype files under {} modified after "
"{}".format(inbox, time.ctime(cutoff_age)))
gt_files_valid = []
for gt_file in filter(GENOTYPE_FILE_RE.match, glob.glob(os.path.join(inbox, "*"))):
if os.stat(gt_file).st_mtime > time.time() - cutoff_age:
gt_files_valid.append(os.path.abspath(gt_file))
if not gt_files_valid:
LOG.info("No genotype files found under {} newer than "
"{}".format(inbox, time.ctime(cutoff_age)))
else:
charon_session = CharonSession()
for gt_file_path in gt_files_valid:
project_samples_dict = \
find_projects_from_samples(parse_samples_from_vcf(gt_file_path))
for project_id, samples in project_samples_dict.iteritems():
LOG.info("Updating project {}...".format(project_id))
for sample in samples:
try:
genotype_status = \
charon_session.sample_get(projectid=project_id,
sampleid=sample).get("genotype_status")
if genotype_status in (None, "NOT_AVAILABLE"):
LOG.info('Updating sample {} genotype_status '
'to "AVAILABLE"...'.format(sample))
charon_session.sample_update(projectid=project_id,
sampleid=sample,
genotype_status="AVAILABLE")
else:
LOG.info('Not updating sample {} genotype_status '
'(already "{}")'.format(sample, genotype_status))
except CharonError as e:
LOG.error('Could not update genotype status to "AVAILABLE" '
'for project/sample "{}/{}": {}'.format(project_id,
sample,
e))
)
except Exception as e:
LOG.error(e.message)
print(e, file=sys.stderr)
LOG.info("Done with organization.")
elif "genotype_project_dirs" in args:
from ngi_pipeline.engines import piper_ngi
genotype_file_path = args.genotype_file
project_obj_list = []
if not args.genotype_project_dirs:
LOG.info("No projects specified; running genotype analysis for all " "samples present in VCF file.")
# User passed only the genotype file; try to determine samples/projects
# from vcf file
projects_samples_dict = find_projects_from_samples(parse_samples_from_vcf(genotype_file_path))
for project_id, samples in projects_samples_dict.iteritems():
try:
path_to_project = locate_project(project_id)
except ValueError:
# Project has not yet been organized from flowcell level
LOG.warn(
'Project "{}" has not yet been organized from '
"flowcell to project level; skipping.".format(project_id)
)
continue
project = recreate_project_from_filesystem(project_dir=path_to_project, restrict_to_samples=samples)
project_obj_list.append(project)
else:
for genotype_project_dir in args.genotype_project_dirs:
LOG.info(
def main(inbox=None,
num_days=14,
genotype_files=None,
config=None,
config_file_path=None):
if genotype_files:
gt_files_valid = [
os.path.abspath(gt_file) for gt_file in genotype_files
]
else:
if not inbox:
try:
inboxes = config["environment"]["flowcell_inbox"]
except (KeyError, TypeError):
raise ValueError(
"No path to delivery inbox specified by argument "
"or in configuration file ({}). Exiting.".format(
config_file_path))
for inbox in inboxes:
inbox = os.path.abspath(inbox)
# Convert to seconds
cutoff_age = time.time() - (int(num_days) * 24 * 60 * 60)
LOG.info("Searching for genotype files under {} modified after "
"{}".format(inbox, time.ctime(cutoff_age)))
gt_files_valid = []
for gt_file in filter(GENOTYPE_FILE_RE.match,
glob.glob(os.path.join(inbox, "*"))):
if os.stat(gt_file).st_mtime > time.time() - cutoff_age:
gt_files_valid.append(os.path.abspath(gt_file))
if not gt_files_valid:
LOG.info("No genotype files found under {} newer than "
"{}".format(inbox, time.ctime(cutoff_age)))
else:
charon_session = CharonSession()
for gt_file_path in gt_files_valid:
project_samples_dict = \
find_projects_from_samples(parse_samples_from_vcf(gt_file_path))
for project_id, samples in project_samples_dict.iteritems():
LOG.info("Updating project {}...".format(project_id))
for sample in samples:
try:
genotype_status = \
charon_session.sample_get(projectid=project_id,
sampleid=sample).get("genotype_status")
if genotype_status in (None, "NOT_AVAILABLE"):
LOG.info('Updating sample {} genotype_status '
'to "AVAILABLE"...'.format(sample))
charon_session.sample_update(
projectid=project_id,
sampleid=sample,
genotype_status="AVAILABLE")
else:
LOG.info('Not updating sample {} genotype_status '
'(already "{}")'.format(
sample, genotype_status))
except CharonError as e:
LOG.error(
'Could not update genotype status to "AVAILABLE" '
'for project/sample "{}/{}": {}'.format(
project_id, sample, e))
except Exception as e:
LOG.error(e.message)
print(e, file=sys.stderr)
LOG.info("Done with organization.")
elif 'genotype_project_dirs' in args:
from ngi_pipeline.engines import piper_ngi
genotype_file_path = args.genotype_file
project_obj_list = []
if not args.genotype_project_dirs:
LOG.info('No projects specified; running genotype analysis for all '
'samples present in VCF file.')
# User passed only the genotype file; try to determine samples/projects
# from vcf file
projects_samples_dict = \
find_projects_from_samples(parse_samples_from_vcf(genotype_file_path))
for project_id, samples in projects_samples_dict.iteritems():
try:
path_to_project = locate_project(project_id)
except ValueError:
# Project has not yet been organized from flowcell level
LOG.warn('Project "{}" has not yet been organized from '
'flowcell to project level; skipping.'.format(project_id))
continue
project = recreate_project_from_filesystem(project_dir=path_to_project,
restrict_to_samples=samples)
project_obj_list.append(project)
else:
for genotype_project_dir in args.genotype_project_dirs:
LOG.info("Starting genotype analysis of project {} with genotype "
"file {}".format(genotype_project_dir, genotype_file_path))
