def test_compute_and_overwrite_existing_column_integer_index(self):
"""Compute new values for an existing column referenced using integer index
"""
tabfile = TabFile('test',self.fp,first_line_is_header=True)
# Check number of columns and header items
self.assertEqual(tabfile.nColumns(),4)
self.assertEqual(tabfile.header(),['chr','start','end','data'])
# Compute new values for data column
tabfile.computeColumn(3,lambda line: line['end'] - line['start'])
self.assertEqual(tabfile.nColumns(),4)
self.assertEqual(tabfile.header(),['chr','start','end','data'])
results = [233,323,4444]
for i in range(len(tabfile)):
self.assertEqual(tabfile[i]['data'],results[i])
def test_compute_and_overwrite_existing_column_integer_index(self):
"""Compute new values for an existing column referenced using integer index
"""
tabfile = TabFile('test', self.fp, first_line_is_header=True)
# Check number of columns and header items
self.assertEqual(tabfile.nColumns(), 4)
self.assertEqual(tabfile.header(), ['chr', 'start', 'end', 'data'])
# Compute new values for data column
tabfile.computeColumn(3, lambda line: line['end'] - line['start'])
self.assertEqual(tabfile.nColumns(), 4)
self.assertEqual(tabfile.header(), ['chr', 'start', 'end', 'data'])
results = [233, 323, 4444]
for i in range(len(tabfile)):
self.assertEqual(tabfile[i]['data'], results[i])
def test_compute_midpoint(self):
"""Compute the midpoint of the start and end columns
"""
tabfile = TabFile('test',self.fp,first_line_is_header=True)
# Check number of columns and header items
self.assertEqual(tabfile.nColumns(),4)
self.assertEqual(tabfile.header(),['chr','start','end','data'])
# Compute midpoint of start and end
tabfile.computeColumn('midpoint',lambda line: (line['end'] + line['start'])/2.0)
self.assertEqual(tabfile.nColumns(),5)
self.assertEqual(tabfile.header(),['chr','start','end','data','midpoint'])
results = [117.5,728.5,3456]
for i in range(len(tabfile)):
self.assertEqual(tabfile[i]['midpoint'],results[i])
def test_compute_midpoint(self):
"""Compute the midpoint of the start and end columns
"""
tabfile = TabFile('test', self.fp, first_line_is_header=True)
# Check number of columns and header items
self.assertEqual(tabfile.nColumns(), 4)
self.assertEqual(tabfile.header(), ['chr', 'start', 'end', 'data'])
# Compute midpoint of start and end
tabfile.computeColumn('midpoint', lambda line:
(line['end'] + line['start']) / 2.0)
self.assertEqual(tabfile.nColumns(), 5)
self.assertEqual(tabfile.header(),
['chr', 'start', 'end', 'data', 'midpoint'])
results = [117.5, 728.5, 3456]
for i in range(len(tabfile)):
self.assertEqual(tabfile[i]['midpoint'], results[i])
