def test_filename(db):
session = Session(db, from_id='testid')
assert session.filename is None
session.filename = 'test'
assert session.filename == 'test'
session = Session(db, from_file='othertest')
assert session.filename == 'othertest'
def test_coordinates(db):
session = Session(db, from_id='testid')
assert session.from_coord == 0
assert session.to_coord == 0
session.from_coord = 23
session.to_coord = 42
assert session.from_coord == 23
assert session.to_coord == 42
def test_region(db):
session = Session(db, from_id='testid')
assert session.region == {}
region = {
'from': 23,
'to': 42,
}
session.region = region
assert session.region == region
def test_last_changed(db, fake_session):
session_id = 12345678
db_session = fake_session(session_id)
session = Session(db, session_id=session_id)
assert session.last_changed == db_session['last_changed']
now = datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S')
session.state = 'done'
assert session.last_changed == now
def test_last_changed_datetime(db, fake_session):
session_id = 12345678
db_session = fake_session(session_id)
session = Session(db, session_id=session_id)
now = datetime.strptime(db_session['last_changed'], '%Y-%m-%d %H:%M:%S')
assert session.last_changed_datetime == now
def test_next(db):
queue = Queue(db, 'test')
session = Session(db, from_id='test123')
queue.submit(session)
key = queue.next()
assert key == session._session_key
key = queue.next()
assert key is None
def test_genome(db):
session = Session(db, from_id='testid')
assert session.genome == {}
genome = {
'locus':
'NC_003888',
'definition':
'Streptomyces coelicolor A3(2) chromosome, complete genome.',
'source':
'Streptomyces coelicolor A3(2)',
'orfs': [
{
'id': 'SCO4711',
'start': 5132708,
'end': 5132995,
'strand': 1,
'gene': 'rpsQ'
},
],
'clusters': [
{
'start':
235986,
'end':
271084,
'name':
"Cluster 3",
'type':
"Lantipeptide",
'description':
"Sanglifehrin_A_biosynthetic_gene_cluster (4% of genes show similarity)"
},
],
}
session.genome = genome
assert session.genome == genome
def test_init_from_id(db, fake_session):
session_id = 123456789
db_session = fake_session(session_id)
session = Session(db, session_id=session_id)
assert session.asid == db_session['asid']
assert session.filename is None
assert session.state == db_session['state']
assert session.error == db_session['error']
assert session.added == db_session['added']
assert session.last_changed == db_session['last_changed']
assert session.genome == {}
assert session.region == {}
assert session.derived is False
assert session.pam == 'GG'
assert session.uniq_size == 13
assert session.full_size == 23
def test_init(db):
now = datetime.utcnow()
session = Session(db, from_id='testid')
session_key = 'crispy:session:{0:039}'.format(session._session_id)
assert session._session_key == session_key
db_session = db.hgetall(session_key)
assert db_session['state'] == 'pending'
assert db_session['asid'] == 'testid'
assert db_session['filename'] == 'None'
assert db_session['added'] == now.strftime('%Y-%m-%d %H:%M:%S')
assert db_session['genome'] == '{}'
assert db_session['region'] == '{}'
assert db_session['from'] == '0'
assert db_session['to'] == '0'
assert db_session['derived'] == 'false'
assert db_session['pam'] == 'GG'
assert db_session['uniq_size'] == '13'
assert db_session['full_size'] == '23'
def test_full_size(db):
session = Session(db, from_id='testid')
assert session.full_size == 23
session.full_size = 17
assert session.full_size == 17
def test_uniq_size(db):
session = Session(db, from_id='testid')
assert session.uniq_size == 13
session.uniq_size = 17
assert session.uniq_size == 17
def test_asid(db):
session = Session(db, from_id='testid')
assert session.asid == 'testid'
session.asid = 'otherid'
assert session.asid == 'otherid'
def test_state(db):
session = Session(db, from_id='testid')
assert session.state == 'pending'
session.state = 'done'
assert session.state == 'done'
def test_init_without_id_or_file(db):
with pytest.raises(ValueError):
session = Session(db)
def test_error(db):
session = Session(db, from_id='testid')
assert session.error == ''
session.error = 'something bad happened'
assert session.error == 'something bad happened'
def test_init_from_invalid_id(db):
with pytest.raises(ValueError):
session = Session(db, session_id=172342)
def test_derived(db):
session = Session(db, from_id='testid')
assert session.derived is False
session.derived = True
assert session.derived is True
def test_init_with_bot_id_and_file(db):
with pytest.raises(ValueError):
session = Session(db, from_id=12345, from_file='test')
def test_derived_raises(db):
session = Session(db, from_id='testid')
with pytest.raises(ValueError):
session.derived = 5
def test_submit(db):
queue = Queue(db, 'test')
assert queue.length == 0
session = Session(db, from_id='test123')
queue.submit(session)
assert queue.length == 1
def test_pam(db):
session = Session(db, from_id='testid')
assert session.pam == 'GG'
session.pam = 'TT'
assert session.pam == 'TT'
def test_repeatability(db):
first = Queue(db, 'test')
second = Queue(db, 'test')
session = Session(db, from_id='test123')
first.submit(session)
assert second.length == 1
def main():
parser = argparse.ArgumentParser(
description='CRISPy sequence preparation service')
parser.add_argument('-d',
'--debug',
dest='debug',
action='store_true',
default=False,
help='print diagnostic messages while running')
parser.add_argument('-q',
'--queue',
dest='queue',
default='redis://127.0.0.1:6379/0',
help='URI of the Queue server')
parser.add_argument(
'-u',
'--upload-dir',
dest='upload_dir',
default='../uploads',
help='Path to directory where the session files are stored')
parser.add_argument('-V',
'--version',
action="version",
version=version.__version__)
args = parser.parse_args()
if args.debug:
logging.basicConfig(format='%(levelname)s %(asctime)s: %(message)s',
level=logging.INFO)
UPLOAD_PATH = path.abspath(args.upload_dir)
db = redis.Redis.from_url(args.queue, decode_responses=True)
queue = Queue(db, 'prepare')
while True:
job_key = queue.next()
if job_key is None:
time.sleep(1)
continue
logging.info('processing {}'.format(job_key))
job_id = int(job_key.split(':')[-1])
job = Session(db, session_id=job_id)
dirname = path.join(UPLOAD_PATH, str(job_id))
if job.asid != 'None':
logging.info('grabbing file for id {}'.format(job.asid))
try:
json_rec, filename = fetch(job.asid, dirname)
job.filename = filename
except requests.HTTPError:
job.state = 'error'
job.error = 'Downloading sequence file failed. Is the ID spelled correctly?'
continue
except Exception:
job.state = 'error'
job.error = 'Failed to load the sequence'
continue
else:
logging.info('parsing uploaded file {}'.format(job.filename))
full_path = path.join(dirname, job.filename)
if job.filename.endswith('.gbk') or job.filename.endswith('.gb'):
try:
rec = list(SeqIO.parse(full_path, 'genbank'))[0]
except (IndexError, ValueError, AssertionError, AttributeError,
UnicodeDecodeError):
job.state = 'error'
job.error = 'Parsing the uploaded file failed. Please check it is a valid GenBank file'
continue
try:
json_rec = genome_json(rec)
except (TypeError):
job.state = 'error'
job.error = 'Invalid input file. Please check the file contains a nucleotide sequence.'
continue
elif job.filename.endswith('.json'):
with open(full_path, 'r') as handle:
as_json = json.load(handle)
json_rec = convert_json(as_json)
gbk_record = json_to_gbk(as_json)
job.filename = 'input.gbk'
SeqIO.write([gbk_record], path.join(dirname, job.filename),
'genbank')
else:
job.state = 'error'
job.error = 'Invalid input file'
continue
job.genome = json_rec
job.state = 'loaded'
logging.info('done with {}'.format(job_key))