def load_samples(manifest_path, database, **kwargs):
manifest = ManifestReader(manifest_path)
# Ensure the sample is marked to be loaded..
if not manifest.marked_for_load():
log.info('Sample not marked for load', extra={
'manifest_path': manifest_path,
})
return
# Ensure the sample section is valid..
if not check_sample_section(manifest):
log.info('Manifest sample section is not valid', extra={
'manifest_path': manifest_path,
})
return
# [sample]
# project = PCGC
# batch = OTHER
# sample = 1-03131
# version = 1
sample_info = manifest.section('sample')
vcf_info = manifest.section('vcf')
# ignore whatever sample is listed in the manifest and scan the vcf
# for samples
vcf_path = os.path.join(os.path.dirname(manifest_path), vcf_info['file'])
with open(vcf_path) as file_obj:
log.debug("opening {0} in load_samples".format(vcf_path))
reader = vcf.Reader(file_obj)
samples = reader.samples
if 'sample' in sample_info:
pretty_names = sample_info['sample'].split(',')
else:
pretty_names = samples
if len(samples) != len(pretty_names):
log.info('Length of comma-delimited samples field in manifest '
'does not match the length of samples in {0}'
.format(vcf_info['file']))
return
# Create the sample (and batch and project if needed)..
num_created = 0
num_skipped = 0
for pretty_name, vcf_sample in zip(pretty_names, samples):
log.debug('Trying to create {0} sample record'.format(vcf_sample))
sample, created = create_sample(sample_name=pretty_name,
vcf_colname=vcf_sample,
batch_name=sample_info['batch'],
project_name=sample_info['project'],
version=sample_info['version'])
log.debug('{0} created'.format(sample))
if created:
num_created += 1
sts.transition(sample, 'Sample Record Created')
else:
num_skipped += 1
manifest = SampleManifest.objects.filter(sample=sample)
# Create a manfiest object for the sample if one does not exist
if created or not manifest.exists():
sample_manifest = SampleManifest(sample=sample)
sample_manifest.load_content(manifest_path)
sample_manifest.save()
sts.transition(sample, 'Sample Manifest Created')
# Publish to channel that this manifest is eligible for processing
# downstream..
if num_created > 0 or kwargs.get('force', False):
SAMPLE_CHANNEL.publish(manifest_path=manifest_path, database=database)
# Returns whether the sample has been created
load_dict = {'created': num_created, 'skipped': num_skipped}
return load_dict
