def test_map_chromosomes(self):
stream = fstream([('chrIV',1),('IV',2),(2780,3),('NC_001136.9',4),('sth',5)], fields=['chr','start'])
assembly = genrep.Assembly('sacCer2')
res = list(map_chromosomes(stream, assembly.chromosomes, keep=True))
expected = [('chrIV',1),('chrIV',2),('chrIV',3),('chrIV',4),('sth',5)]
self.assertListEqual(res, expected)
# keep=False
stream = fstream([('chrIV',1),('IV',2),(2780,3),('NC_001136.9',4),('sth',5)], fields=['chr','start'])
res = list(map_chromosomes(stream, assembly.chromosomes, keep=False))
self.assertListEqual(res, expected[:-1])
def test_map_chromosomes(self):
stream = fstream([('chrIV', 1), ('IV', 2), (2780, 3),
('NC_001136.9', 4), ('sth', 5)],
fields=['chr', 'start'])
assembly = genrep.Assembly('sacCer2')
res = list(map_chromosomes(stream, assembly.chromosomes, keep=True))
expected = [('chrIV', 1), ('chrIV', 2), ('chrIV', 3), ('chrIV', 4),
('sth', 5)]
self.assertListEqual(res, expected)
# keep=False
stream = fstream([('chrIV', 1), ('IV', 2), (2780, 3),
('NC_001136.9', 4), ('sth', 5)],
fields=['chr', 'start'])
res = list(map_chromosomes(stream, assembly.chromosomes, keep=False))
self.assertListEqual(res, expected[:-1])
def convert_junc_file(self, filename):
"""Convert a .junc SOAPsplice output file to bed format. Return the file name.
:param filename: (str) name of the .junc file to convert.
"""
t = track(filename, format='txt', fields=['chr','start','end','strand','score'],
chrmeta=self.assembly.chrmeta)
stream = t.read()
# Translate chromosome names
s1 = map_chromosomes(stream, self.assembly.chromosomes)
# Add junction IDs
s2 = duplicate(s1,'strand','name')
C = itertools.count()
s3 = apply(s2,'name', lambda x: 'junction'+str(C.next()))
# Convert to bed format
outfile = unique_filename_in()
bed = outfile + '.bed'
out = track(bed, fields=s3.fields, chrmeta=self.assembly.chrmeta)
out.write(s3)
return bed
def convert_junc_file(self, filename):
"""Convert a .junc SOAPsplice output file to bed format. Return the file name.
:param filename: (str) name of the .junc file to convert.
"""
t = track(filename,
format='txt',
fields=['chr', 'start', 'end', 'strand', 'score'],
chrmeta=self.assembly.chrmeta)
stream = t.read()
# Translate chromosome names
s1 = map_chromosomes(stream, self.assembly.chromosomes)
# Add junction IDs
s2 = duplicate(s1, 'strand', 'name')
C = itertools.count()
s3 = apply(s2, 'name', lambda x: 'junction' + str(C.next()))
# Convert to bed format
outfile = unique_filename_in()
bed = outfile + '.bed'
out = track(bed, fields=s3.fields, chrmeta=self.assembly.chrmeta)
out.write(s3)
return bed
