def test_slice_mapping_file(self):
header, metadata = parse_mapping_file(StringIO(test_mapping))
table = Table(array([[1, 2], [4, 5]]), ['x', 'y'], ['a', 'c'])
exp = ["a\t1\t123123", "c\tpoop\tdoesn't matter"]
obs = slice_mapping_file(table, metadata)
self.assertEqual(obs, exp)
def test_slice_mapping_file(self):
header, metadata = parse_mapping_file(StringIO(test_mapping))
table = Table(array([[1,2],[4,5]]),
['x','y'],
['a','c'])
exp = ["a\t1\t123123", "c\tpoop\tdoesn't matter"]
obs = slice_mapping_file(table, metadata)
self.assertEqual(obs,exp)
def test_parse_mapping_file(self):
exp = ("#SampleIDs\tfoo\tbar", [['a','1\t123123'],
['b','yy\txxx'],
['c',"poop\tdoesn't matter"]])
obs = parse_mapping_file(StringIO(test_mapping))
self.assertEqual(obs, exp)
def test_parse_mapping_file(self):
exp = ("#SampleIDs\tfoo\tbar", [['a', '1\t123123'], ['b', 'yy\txxx'],
['c', "poop\tdoesn't matter"]])
obs = parse_mapping_file(StringIO(test_mapping))
self.assertEqual(obs, exp)
from sys import argv, exit
__author__ = "Daniel McDonald"
__copyright__ = "Copyright 2013, The American Gut Project"
__credits__ = ["Daniel McDonald"]
__license__ = "BSD"
__version__ = "unversioned"
__maintainer__ = "Daniel McDonald"
__email__ = "*****@*****.**"
if __name__ == '__main__':
if len(argv) != 4:
print "usage: python %s mappingfile table output" % argv[0]
exit(1)
header, mapping_file = parse_mapping_file(open(argv[1]))
table = load_table(argv[2])
if not verify_subset(table, mapping_file):
print "****"
print set([i[0] for i in mapping_file]) - set(table.ids())
print set(table.ids()) - set([i[0] for i in mapping_file])
raise ValueError("The table is not a subset of the mapping file!")
sliced = slice_mapping_file(table, mapping_file)
f = open(argv[3], 'w')
f.write(header)
f.write('\n')
f.write('\n'.join(sliced))
__author__ = "Daniel McDonald"
__copyright__ = "Copyright 2013, The American Gut Project"
__credits__ = ["Daniel McDonald"]
__license__ = "BSD"
__version__ = "unversioned"
__maintainer__ = "Daniel McDonald"
__email__ = "*****@*****.**"
if __name__ == "__main__":
if len(argv) != 4:
print "usage: python %s mappingfile table output" % argv[0]
exit(1)
header, mapping_file = parse_mapping_file(open(argv[1]))
table = load_table(argv[2])
if not verify_subset(table, mapping_file):
print "****"
print set([i[0] for i in mapping_file]) - set(table.ids())
print set(table.ids()) - set([i[0] for i in mapping_file])
raise ValueError("The table is not a subset of the mapping file!")
sliced = slice_mapping_file(table, mapping_file)
f = open(argv[3], "w")
f.write(header)
f.write("\n")
f.write("\n".join(sliced))
