def test_get_custom_prority():
"""docstring for test_get_chr_prority"""
assert get_chromosome_priority(chrom='AHA_1',
chrom_dict={
'AHA_1': 2,
'AHA_2': 3
}) == 2
def run(self):
"""Starts the printing"""
# Print the results to a temporary file:
number_of_finished = 0
proc_name = self.name
self.logger.info(('{0}: starting'.format(proc_name)))
if self.outfile:
if isinstance(self.outfile, str):
self.outfile = open(self.outfile, 'w+', encoding="utf-8")
while True:
# A task is a variant dictionary
self.logger.debug(('{0} fetching next variant'.format(proc_name)))
variant = self.task_queue.get()
if self.task_queue.full():
self.logger.warning('Variant queue full')
if variant is None:
self.logger.info('All variants printed.')
if self.outfile:
self.outfile.close()
break
self.logger.debug("Printing variant {0}".format(variant.get('variant_id', 'unknown')))
priority = None
if self.mode == 'chromosome':
priority = get_chromosome_priority(variant['CHROM'])
elif self.mode == 'score':
priority = get_rank_score(variant_dict=variant)
print_variant(variant_dict=variant, header_line=self.header,
priority=priority, outfile=self.outfile,
silent=self.silent)
return
def run(self):
"""Starts the printing"""
# Print the results to a temporary file:
number_of_finished = 0
proc_name = self.name
self.logger.info(('{0}: starting'.format(proc_name)))
if self.outfile:
if isinstance(self.outfile, str):
self.outfile = open(self.outfile, 'w+', encoding="utf-8-sig")
while True:
# A task is a variant dictionary
self.logger.debug(('{0} fetching next variant'.format(proc_name)))
variant = self.task_queue.get()
if self.task_queue.full():
self.logger.warning('Variant queue full')
if variant is None:
self.logger.info('All variants printed.')
if self.outfile:
self.outfile.close()
break
self.logger.debug("Printing variant {0}".format(variant.get('variant_id', 'unknown')))
priority = None
if self.mode == 'chromosome':
priority = get_chromosome_priority(variant['CHROM'])
elif self.mode == 'score':
priority = get_rank_score(variant_dict=variant)
print_variant(variant_dict=variant, header_line=self.header,
priority=priority, outfile=self.outfile,
silent=self.silent)
return
def sort(variant_file, outfile, family_id, silent, position, temp_dir):
"""
Sort a VCF file based on rank score.
"""
logger = logging.getLogger(__name__)
head = HeaderParser()
logger.info("Running GENMOD sort version {0}".format(__version__))
start = datetime.now()
# Create a temporary variant file for sorting
logger.debug("Creating temporary file for sorting")
if temp_dir:
temp_file = NamedTemporaryFile(delete=False, dir=temp_dir)
else:
temp_file = NamedTemporaryFile(delete=False)
temp_file.close()
# Open the temp file with codecs
temp_file_handle = open(
temp_file.name,
mode='w',
encoding='utf-8',
errors='replace'
)
logger.debug("Temp file created")
logger.info("Printing variants to temp file")
nr_variants = 0
# Print the variants with rank score in first column
for line in variant_file:
line = line.rstrip()
if line.startswith('#'):
if line.startswith('##'):
head.parse_meta_data(line)
else:
head.parse_header_line(line)
else:
nr_variants += 1
priority = '0'
if position:
chrom = line.split()[0]
priority = get_chromosome_priority(chrom)
else:
priority = get_rank_score(line)
print_variant(
variant_line=line,
priority=priority,
outfile=temp_file_handle
)
temp_file_handle.close()
logger.info("Variants printed to temp file")
logger.info("Nr or variants in VCF file: {0}".format(nr_variants))
sort_mode = 'rank'
if position:
sort_mode = 'chromosome'
logger.info("Sorting variants")
sort_variants(
infile = temp_file.name,
mode=sort_mode
)
logger.info("Variants sorted")
logger.debug("Printing headers")
print_headers(
head = head,
outfile = outfile,
silent=silent
)
logger.debug("Headers printed")
logger.info("Printing variants")
with open(temp_file.name, mode='r', encoding='utf-8', errors='replace') as f:
for variant_line in f:
print_variant(
variant_line = variant_line,
outfile = outfile,
mode = 'modified',
silent=False
)
logger.debug("Variants printed")
logger.info("Removing temp file")
os.remove(temp_file.name)
logger.debug("Temp file removed")
logger.info("Sorting done, time for sorting: {0}".format(datetime.now()-start))
def sort(variant_file, outfile, family_id, silent, position, temp_dir):
"""
Sort a VCF file based on rank score.
"""
head = HeaderParser()
variant_file = get_file_handle(variant_file)
logger.info("Running GENMOD sort version {0}".format(__version__))
start = datetime.now()
# Create a temporary variant file for sorting
logger.debug("Creating temporary file for sorting")
if temp_dir:
temp_file = NamedTemporaryFile(delete=False, dir=temp_dir)
else:
temp_file = NamedTemporaryFile(delete=False)
temp_file.close()
# Open the temp file with codecs
temp_file_handle = open(temp_file.name,
mode='w',
encoding='utf-8',
errors='replace')
logger.debug("Temp file created")
logger.info("Printing variants to temp file")
nr_variants = 0
# Print the variants with rank score in first column
for line in variant_file:
line = line.rstrip()
if line.startswith('#'):
if line.startswith('##'):
head.parse_meta_data(line)
else:
head.parse_header_line(line)
else:
nr_variants += 1
priority = '0'
if position:
chrom = line.split()[0]
priority = get_chromosome_priority(chrom)
else:
priority = get_rank_score(line)
print_variant(variant_line=line,
priority=priority,
outfile=temp_file_handle)
temp_file_handle.close()
logger.info("Variants printed to temp file")
logger.info("Nr or variants in VCF file: {0}".format(nr_variants))
sort_mode = 'rank'
if nr_variants == 0:
logger.debug("Printing headers")
print_headers(head=head, outfile=outfile, silent=silent)
sys.exit(0)
if position:
sort_mode = 'chromosome'
logger.info("Sorting variants")
sort_variants(infile=temp_file.name, mode=sort_mode)
logger.info("Variants sorted")
logger.debug("Printing headers")
print_headers(head=head, outfile=outfile, silent=silent)
logger.debug("Headers printed")
logger.info("Printing variants")
with open(temp_file.name, mode='r', encoding='utf-8',
errors='replace') as f:
for variant_line in f:
print_variant(variant_line=variant_line,
outfile=outfile,
mode='modified',
silent=False)
logger.debug("Variants printed")
logger.info("Removing temp file")
os.remove(temp_file.name)
logger.debug("Temp file removed")
logger.info("Sorting done, time for sorting: {0}".format(datetime.now() -
start))
def test_get_X_priority():
"""docstring for test_get_X_priority"""
assert get_chromosome_priority(chrom='X', chrom_dict={}) == '23'
def test_get_custom_prority():
"""docstring for test_get_chr_prority"""
assert get_chromosome_priority(chrom='AHA_1', chrom_dict={'AHA_1':2, 'AHA_2':3}) == '2'
def test_get_chr_prority():
"""docstring for test_get_chr_prority"""
assert get_chromosome_priority(chrom='chr1', chrom_dict={}) == '1'
def test_get_OTHER_priority():
"""docstring for test_get_MT_priority"""
assert get_chromosome_priority(chrom='GL37', chrom_dict={}) == '26'
def test_get_X_priority():
"""docstring for test_get_X_priority"""
assert get_chromosome_priority(chrom='X', chrom_dict={}) == '23'
def test_get_chr_prority():
"""docstring for test_get_chr_prority"""
assert get_chromosome_priority(chrom='chr1', chrom_dict={}) == '1'
def test_get_OTHER_priority():
"""docstring for test_get_MT_priority"""
assert get_chromosome_priority(chrom='GL37', chrom_dict={}) == '26'
