def test_error_on_too_many_states():
nex = NexusReader.from_string("""
Begin data;
Dimensions ntax=1 nchar=30;
Format datatype=standard symbols="01" gap=-;
Matrix
A 111111111111111111111111111111
;""")
with pytest.raises(ValueError):
multistatise(nex)
def test_error_on_too_many_states(self):
self.nex = NexusReader()
self.nex.read_string("""
Begin data;
Dimensions ntax=1 nchar=30;
Format datatype=standard symbols="01" gap=-;
Matrix
A 111111111111111111111111111111
;""")
with self.assertRaises(ValueError):
multistatise(self.nex)
def nex():
res = NexusReader.from_string("""
Begin data;
Dimensions ntax=4 nchar=4;
Format datatype=standard symbols="01" gap=-;
Matrix
Harry 1000
Simon 0100
Betty 0010
Louise 0001
;""")
return multistatise(res)
def setUp(self):
self.nex = NexusReader()
self.nex.read_string("""
Begin data;
Dimensions ntax=4 nchar=4;
Format datatype=standard symbols="01" gap=-;
Matrix
Harry 1000
Simon 0100
Betty 0010
Louise 0001
;""")
self.nex = multistatise(self.nex)
def setUp(self):
self.nex = NexusReader()
self.nex.read_string("""
Begin data;
Dimensions ntax=4 nchar=4;
Format datatype=standard symbols="01" gap=-;
Matrix
Harry 1000
Simon 0100
Betty 0010
Louise 0001
;""")
self.nex = multistatise(self.nex)
def test_regression_include_invisible_taxa(self):
"""Include taxa that have no entries"""
data = """
#NEXUS
BEGIN DATA;
DIMENSIONS NTAX=15 NCHAR=7;
FORMAT DATATYPE=STANDARD MISSING=? GAP=- INTERLEAVE=YES;
MATRIX
Gertrude 0000001
Debbie 0001000
Zarathrustra 0000000
Christie 0010000
Benny 0100000
Bertha 0100000
Craig 0010000
Fannie-May 0000010
Charles 0010000
Annik 1000000
Frank 0000010
Amber 1000000
Andreea 1000000
Edward 0000100
Donald 0001000
;
END;
"""
nex = NexusReader()
nex.read_string(data)
msnex = multistatise(nex)
for taxon, sites in msnex.data.matrix.items():
if taxon[0] == 'Z':
continue # will check later
# first letter of taxa name is the expected character state
assert taxon[0] == sites[0], \
"%s should be %s not %s" % (taxon, taxon[0], sites[0])
# deal with completely missing taxa
assert 'Zarathrustra' in msnex.data.matrix
assert msnex.data.matrix['Zarathrustra'][0] == '?'
def test_regression_include_invisible_taxa(self):
"""Include taxa that have no entries"""
data = """
#NEXUS
BEGIN DATA;
DIMENSIONS NTAX=15 NCHAR=7;
FORMAT DATATYPE=STANDARD MISSING=? GAP=- INTERLEAVE=YES;
MATRIX
Gertrude 0000001
Debbie 0001000
Zarathrustra 0000000
Christie 0010000
Benny 0100000
Bertha 0100000
Craig 0010000
Fannie-May 0000010
Charles 0010000
Annik 1000000
Frank 0000010
Amber 1000000
Andreea 1000000
Edward 0000100
Donald 0001000
;
END;
"""
nex = NexusReader()
nex.read_string(data)
msnex = multistatise(nex)
for taxon, sites in msnex.data.matrix.items():
if taxon[0] == 'Z':
continue # will check later
# first letter of taxa name is the expected character state
assert taxon[0] == sites[0], \
"%s should be %s not %s" % (taxon, taxon[0], sites[0])
# deal with completely missing taxa
assert 'Zarathrustra' in msnex.data.matrix
assert msnex.data.matrix['Zarathrustra'][0] == '?'
usage="usage: %prog [-o output.nex] nex1.nex nex2.nex ... nexN.nex")
parser.add_option("-o",
"--output",
dest="output",
action="store",
default=None,
type="string",
help="output nexus file")
options, nexuslist = parser.parse_args()
if len(nexuslist) < 1:
print(__doc__)
parser.print_help()
sys.exit()
if options.output is not None:
outfile = options.output
else:
outfile = 'multistate.nex'
nexuslist2 = []
for nfile in nexuslist:
n = NexusReader(nfile)
n = multistatise.multistatise(n)
nexuslist2.append(n)
out = combine_nexuses(nexuslist2)
out.write_to_file(outfile, charblock=True, interleave=False)
print("Written to %s" % outfile)
def run(args):
nexuslist = [
multistatise.multistatise(n) for n in get_reader(args, many=True)
]
write_output(combine_nexuses(nexuslist), args)