def __init__(self, screen_file):
"""
Create a new FastqscreenData instance
"""
TabFile.__init__(self,
column_names=(
'Library',
'%Unmapped',
'%One_hit_one_library',
'%Multiple_hits_one_library',
'%One_hit_multiple_libraries',
'%Multiple_hits_multiple_libraries',
))
self._screen_file = os.path.abspath(screen_file)
self._version = None
self._no_hits = None
# Read in data
with open(self._screen_file, 'r') as fp:
for line in fp:
line = line.strip()
if line.startswith('#Fastq_screen version:'):
self._version = line.split()[2]
continue
elif line.startswith('Library') or line.startswith('Genome'):
tabfile = TabFile(column_names=line.split())
continue
elif line.startswith('%Hit_no_libraries:') or \
line.startswith('%Hit_no_genomes:'):
self._no_hits = float(line.split()[-1])
continue
elif not line or \
line.startswith('#') or \
line.startswith('%'):
continue
tabfile.append(tabdata=line)
# Handle different terminology for different versions
if tabfile.header()[0] == 'Library':
library = 'Library'
unmapped = '%Unmapped'
one_hit_one_library = '%One_hit_one_library'
multiple_hits_one_library = '%Multiple_hits_one_library'
one_hit_multiple_libraries = '%One_hit_multiple_libraries'
multiple_hits_multiple_libraries = '%Multiple_hits_multiple_libraries'
elif tabfile.header()[0] == 'Genome':
library = 'Genome'
unmapped = '%Unmapped'
one_hit_one_library = '%One_hit_one_genome'
multiple_hits_one_library = '%Multiple_hits_one_genome'
one_hit_multiple_libraries = '%One_hit_multiple_genomes'
multiple_hits_multiple_libraries = '%Multiple_hits_multiple_genomes'
# Copy data to main object
for line in tabfile:
data = [
line[library], line[unmapped], line[one_hit_one_library],
line[multiple_hits_one_library],
line[one_hit_multiple_libraries],
line[multiple_hits_multiple_libraries]
]
self.append(data=data)
def __init__(self,f,multiline=False):
"""
Create a new MetricsSummary instance
Arguments:
f (str): path to the 'metrics_summary.csv' file
multiline (bool): if True then expect multiple lines
of data (default is to expect a single line of
data)
"""
# Multi-line summary?
self._multiline = multiline
# Read in data from the file
with open(f,'rt') as fp:
s = fp.read()
self._data = dict()
s = s.strip().split('\n')
if not self._multiline and len(s) != 2:
raise Exception("%s: MetricsSummary expects 2 lines (or specify "
"multi-line mode)" % f)
# Set up the tabfile instance
TabFile.__init__(self,
column_names=self._tokenise(s[0]),
delimiter=',')
# Add the data
for line in s[1:]:
self.append(data=self._tokenise(line))
def __init__(self,screen_file):
"""
Create a new FastqscreenData instance
"""
TabFile.__init__(self,
column_names=('Library',
'%Unmapped',
'%One_hit_one_library',
'%Multiple_hits_one_library',
'%One_hit_multiple_libraries',
'%Multiple_hits_multiple_libraries',))
self._screen_file = os.path.abspath(screen_file)
self._version = None
self._no_hits = None
# Read in data
with open(self._screen_file,'r') as fp:
for line in fp:
line = line.strip()
if line.startswith('#Fastq_screen version:'):
self._version = line.split()[2]
continue
elif line.startswith('Library') or line.startswith('Genome'):
tabfile = TabFile(column_names=line.split())
continue
elif line.startswith('%Hit_no_libraries:') or \
line.startswith('%Hit_no_genomes:'):
self._no_hits = float(line.split()[-1])
continue
elif not line or \
line.startswith('#') or \
line.startswith('%'):
continue
tabfile.append(tabdata=line)
# Handle different terminology for different versions
if tabfile.header()[0] == 'Library':
library = 'Library'
unmapped = '%Unmapped'
one_hit_one_library = '%One_hit_one_library'
multiple_hits_one_library = '%Multiple_hits_one_library'
one_hit_multiple_libraries = '%One_hit_multiple_libraries'
multiple_hits_multiple_libraries = '%Multiple_hits_multiple_libraries'
elif tabfile.header()[0] == 'Genome':
library = 'Genome'
unmapped = '%Unmapped'
one_hit_one_library = '%One_hit_one_genome'
multiple_hits_one_library = '%Multiple_hits_one_genome'
one_hit_multiple_libraries = '%One_hit_multiple_genomes'
multiple_hits_multiple_libraries = '%Multiple_hits_multiple_genomes'
# Copy data to main object
for line in tabfile:
data = [line[library],
line[unmapped],
line[one_hit_one_library],
line[multiple_hits_one_library],
line[one_hit_multiple_libraries],
line[multiple_hits_multiple_libraries]]
self.append(data=data)
def __init__(self,f):
"""
Create a new AtacSummary instance
Arguments:
f (str): path to the 'summary.csv' file
"""
TabFile.__init__(self,
filen=f,
first_line_is_header=True,
delimiter=',')
def __init__(self, summary_file=None):
"""
Create a new FastqcSummary instance
"""
TabFile.__init__(self, column_names=("Status", "Module", "File"))
if summary_file:
summary_file = os.path.abspath(summary_file)
with open(summary_file, "r") as fp:
for line in fp:
line = line.strip()
self.append(tabdata=line)
self._summary_file = summary_file
def __init__(self, summary_file=None):
"""
Create a new FastqcSummary instance
"""
TabFile.__init__(self, column_names=(
'Status',
'Module',
'File',
))
if summary_file:
summary_file = os.path.abspath(summary_file)
with open(summary_file, 'r') as fp:
for line in fp:
line = line.strip()
self.append(tabdata=line)
self._summary_file = summary_file
def __init__(self,screen_file):
"""
Create a new FastqscreenData instance
"""
TabFile.__init__(self,
column_names=('Library',
'%Unmapped',
'%One_hit_one_library',
'%Multiple_hits_one_library',
'%One_hit_multiple_libraries',
'%Multiple_hits_multiple_libraries',))
self._screen_file = os.path.abspath(screen_file)
self._version = None
self._no_hits = None
# Read in data
with open(self._screen_file,'r') as fp:
for line in fp:
line = line.strip()
if line.startswith('#Fastq_screen version:'):
self._version = line.split()[2]
continue
elif line.startswith('Library'):
tabfile = TabFile(column_names=line.split())
continue
elif line.startswith('%Hit_no_libraries:'):
self._no_hits = float(line.split()[-1])
continue
elif not line or \
line.startswith('#') or \
line.startswith('%'):
continue
tabfile.append(tabdata=line)
# Move data to main object
for line in tabfile:
data = []
for col in self.header():
data.append(line[col])
self.append(data=data)
