def open_json(self, x=None):
idxs = self.jlist.curselection()
if not idxs:
lo.warning('cannot open JsonExplorer : idxs='+str(idxs))
return
jpath = self.jpaths[int(idxs[0])]
self.do_open_json(jpath)
def update(args):
if args.fastq:
lo.warning('re-reading of hits not currently implemented')
data = json.load(file(args.json))
testsuite_paths = discover_testsuites(args.testsuite_directory or [])
testsuites = {}
update_testsuites(testsuites, data['info']['testsuites'], testsuite_paths)
analyser = analyse.Analyser()
analyser.decode(testsuites, data)
analyser.update_testsuites()
# save results back to .json
data = analyser.encode(hits = analyser.hits is not None)
j = codecs.open(args.json, 'w', 'utf-8')
lo.info('re-wrote results to file ' + args.json)
json.dump(data, j, indent=2)
def scan_illustrate(self, MTBC_fastq, scan_params=[]):
ntf = tempfile.NamedTemporaryFile(delete=False)
lo.setLevel(logging.WARNING)
try:
t0 = time.time()
out, err = self.main(['scan', '-l', 'MTBC', '-f'] + scan_params + [
MTBC_fastq, ntf.name])
if time.time() - t0 > 10:
lo.warning('scanning of %s took %.2fs' % (
os.path.basename(MTBC_fastq), time.time() - t0))
out, err = self.main(['illustrate', '-r',
ntf.name])
finally:
lo.setLevel(logging.INFO)
ntf.close()
os.remove(ntf.name)
return out, err
def test_load_testsuites(self):
def get_testsuites(out):
line = [line for line in out.split('\n')
if line.startswith('testsuites=')][0]
return set([testsuite
for testsuite in line[line.index('=')+1:].split(',')
if testsuite])
def name_and_version(testsuites):
return set([testsuite[:testsuite.find('[')]
for testsuite in testsuites])
def name_only(testsuites):
return set([testsuite[:testsuite.find('-')]
for testsuite in testsuites])
cwd = os.getcwd()
out, err = self.main(['info'])
assert get_testsuites(out) == set()
# suppress "loaded testsuite" messages
lo.setLevel(logging.WARNING)
MTBC_testsuites = set(['MTBC/phylo', 'MTBC/resistance', 'MTBC/spoligo'])
# select testsuite
out, err = self.main(['info', '-l', 'MTBC/phylo'])
assert name_only(get_testsuites(out)) == set(['MTBC/phylo'])
# select group
out, err = self.main(['info', '-l', 'MTBC'])
assert name_only(get_testsuites(out)) == MTBC_testsuites
# select testsuite by filename
path = os.path.join(testsuites_alt, 'MTBC', 'phylo.py')
out, err = self.main(['info', '-l', path])
assert name_and_version(get_testsuites(out)) == set(['MTBC/phylo-0.0'])
# override testsuite directory using switch
out, err = self.main(['info', '-l', 'MTBC/phylo'])
assert name_and_version(get_testsuites(out)) != set(['MTBC/phylo-0.0'])
out, err = self.main(['-t', testsuites_alt, 'info', '-l', 'MTBC/phylo'])
assert name_and_version(get_testsuites(out)) == set(['MTBC/phylo-0.0'])
# override testsuite directory using KVARQ_TESTSUITES
os.environ['KVARQ_TESTSUITES'] = testsuites_alt
out, err = self.main(['info', '-l', 'MTBC/phylo'])
assert name_and_version(get_testsuites(out)) == set(['MTBC/phylo-0.0'])
del os.environ['KVARQ_TESTSUITES']
# time load all
t0 = time.time()
out, err = self.main(['info', '-L'])
assert len(name_only(get_testsuites(out))) > 4
if time.time() - t0 > 2:
lo.warning('loading all testsuites takes %.2f' % (time.time() - t0))
lo.setLevel(logging.INFO)
os.chdir(cwd)
def __init__(self, fname, variant=None, fd=None, paired=False, quiet=False):
'''
open ``.fastq`` or ``.fastq.gz`` file and determine its
variant (setting attribute ``.Azero`` accordingly)
:param fname: name of file to open
:param variant: specify one of ``.vendor_variants`` -- if none
is specified, then the PHRED score of the fastq file is
analyzed and
:param fd: specify a file descriptor to use instead of
opening ``fname``
:param paired: include second file in a paired set if it is
available (i.e. specify "file_1.fastq" as input file and
"file_2.fastq" will be included in functions ``.filesize()``
and ``.filenames()``)
'''
self.fname = fname
if fd:
self.fd = fd
else:
self.fd = None
if self.fname.endswith('.fastq.gz'):
self.gz = True
if not self.fd:
self.fd = gzip.GzipFile(self.fname, 'rb')
elif self.fname.endswith('.fastq'):
self.gz = False
if not self.fd:
self.fd = open(self.fname, 'rb')
else:
raise FastqFileFormatException(
'fastq file must have extension ".fastq" or ".fastq.gz"')
# save second name of base if exists
self.fname2 = None
if paired:
base = fname[:fname.rindex('.fastq')]
if base[-2:] == '_1':
fname2 = base[:-2] + '_2' + fname[fname.rindex('.fastq'):]
if os.path.exists(fname2):
lo.info('including paired file "%s"' % fname2)
self.fname2 = fname2
if sum(self.filesizes()) == 0:
raise FastqFileFormatException('cannot scan empty file')
# scan some records
min_pos, max_pos = self.min_max_score_check_file()
lo.debug('min_pos=%d max_pos=%d' % (min_pos, max_pos))
if variant and variant not in self.vendor_variants:
raise FastqFileFormatException(
'unknown vendor variant "%s"' % variant)
# create list of variants compatible with PHRED scores
variants = []
dQs = []
for name, vendor_variant in Fastq.vendor_variants.items():
if ((min_pos - vendor_variant.dQ) in vendor_variant.Qrange
and (max_pos - vendor_variant.dQ) in vendor_variant.Qrange):
dQs.append(vendor_variant.dQ)
variants.append(name)
if variant is None:
# set variant from guesses
if not variants:
raise FastqFileFormatException(
'could not find any suitable fastq vendor variant')
if len(set(dQs)) > 1:
raise FastqFileFormatException(
'cannot determine dQ with guessed vendor variants "%s"'
% str(variants))
self.variants = variants
self.dQ = dQs[0]
else:
# check specified variant
if variant not in variants:
lo.warning('specified vendor variant "%s" seems not to be '
'compatible with file' % variant)
self.variants = [variant]
self.dQ = self.vendor_variants[variant].dQ
self.Azero = self.ASCII[self.dQ]
# estimate readlength/records_approx
self.fd.seek(0)
lines = [self.fd.readline() for i in range(4)]
self.readlength = len(lines[1].strip('\r\n'))
if self.gz:
self.records_approx = None
else:
self.records_approx = os.path.getsize(self.fname) / len(''.join(lines))
if self.fname2 is not None:
self.records_approx *= 2
# output some infos
if not quiet:
if self.gz:
lo.info('gzipped fastq : readlength=? records_approx=? dQ=%d variants=%s' % (
self.dQ, str(self.variants)))
else:
lo.info('fastq : readlength=%d records_approx=%d dQ=%d variants=%s' % (
self.readlength, self.records_approx, self.dQ, str(self.variants)))
