def test_split__three_groups():
msa = {"seq1" : "ACGCAT", "seq2" : "GAGTGA"}
expected = {"1" : {"seq1" : "AC", "seq2" : "GT"},
"2" : {"seq1" : "CA", "seq2" : "AG"},
"3" : {"seq1" : "GT", "seq2" : "GA"}}
with RequiredCall(_VALIDATION_PATH, args = [msa]):
assert_equal(split_msa(msa, "123"), expected)
with RequiredCall(_VALIDATION_PATH, args = [msa]):
assert_equal(split_msa(msa), expected)
def test_split__partial_group():
msa = {"seq1" : "ACGCA", "seq2" : "GAGTG"}
expected = {"1" : {"seq1" : "AC", "seq2" : "GT"},
"2" : {"seq1" : "CA", "seq2" : "AG"},
"3" : {"seq1" : "G", "seq2" : "G"}}
with RequiredCall(_VALIDATION_PATH, args = [msa]):
assert_equal(split_msa(msa), expected)
def test_split__empty_group():
msa = {"seq1" : "AC", "seq2" : "GA"}
expected = {"1" : {"seq1" : "A", "seq2" : "G"},
"2" : {"seq1" : "C", "seq2" : "A"},
"3" : {"seq1" : "", "seq2" : ""}}
with RequiredCall(_VALIDATION_PATH, args = [msa]):
assert_equal(split_msa(msa), expected)
def test_split__three_groups():
msa = {"seq1": "ACGCAT", "seq2": "GAGTGA"}
expected = {
"1": {
"seq1": "AC",
"seq2": "GT"
},
"2": {
"seq1": "CA",
"seq2": "AG"
},
"3": {
"seq1": "GT",
"seq2": "GA"
}
}
with RequiredCall(_VALIDATION_PATH, args=[msa]):
assert_equal(split_msa(msa, "123"), expected)
with RequiredCall(_VALIDATION_PATH, args=[msa]):
assert_equal(split_msa(msa), expected)
def test_split_msa__two_groups():
msa = {"seq1": "ACGCAT", "seq2": "GAGTGA"}
expected = {
"1": {
"seq1": "ACCA",
"seq2": "GATG"
},
"2": {
"seq1": "GT",
"seq2": "GA"
}
}
with RequiredCall(_VALIDATION_PATH, args=[msa]):
assert_equal(split_msa(msa, "112"), expected)
def test_split__partial_group():
msa = {"seq1": "ACGCA", "seq2": "GAGTG"}
expected = {
"1": {
"seq1": "AC",
"seq2": "GT"
},
"2": {
"seq1": "CA",
"seq2": "AG"
},
"3": {
"seq1": "G",
"seq2": "G"
}
}
with RequiredCall(_VALIDATION_PATH, args=[msa]):
assert_equal(split_msa(msa), expected)
def test_split__empty_group():
msa = {"seq1": "AC", "seq2": "GA"}
expected = {
"1": {
"seq1": "A",
"seq2": "G"
},
"2": {
"seq1": "C",
"seq2": "A"
},
"3": {
"seq1": "",
"seq2": ""
}
}
with RequiredCall(_VALIDATION_PATH, args=[msa]):
assert_equal(split_msa(msa), expected)
def _run(self, _config, temp):
msas = []
for filename in sorted(self._infiles):
split_by = self._infiles[filename].get("partition_by", self._part_by)
for (key, msa) in sorted(split_msa(read_msa(filename), split_by).items()):
for excluded_group in self._excluded:
msa.pop(excluded_group)
msas.append(("%s_%s" % (self._infiles[filename]["name"], key), msa))
msa = join_msa(*(msa for (_, msa) in msas))
with open(reroot_path(temp, self._out_prefix + ".phy"), "w") as output:
output.write(interleaved_phy(msa, add_flag = self._add_flag))
with open(reroot_path(temp, self._out_prefix + ".partitions"), "w") as output:
end = 0
for (name, msa) in msas:
length = len(msa.itervalues().next())
output.write("DNA, %s = %i-%i\n" % (name, end + 1, end + length))
end += length
def test_split_msa__no_split_by():
split_msa({"seq1": "ACG", "seq2": "GAT"}, split_by="")
def test_split_msa__empty_msa():
split_msa({})
def test_split_msa__single_group():
msa = {"seq1": "ACGCAT", "seq2": "GAGTGA"}
expected = {'1': {"seq1": "ACGCAT", "seq2": "GAGTGA"}}
with RequiredCall(_VALIDATION_PATH, args=[msa]):
assert_equal(split_msa(msa, "111"), expected)
def test_split_msa__no_split_by():
split_msa({"seq1" : "ACG", "seq2" : "GAT"}, split_by = "")
def test_split_msa__empty_msa():
split_msa({})
def test_split_msa__two_groups():
msa = {"seq1" : "ACGCAT", "seq2" : "GAGTGA"}
expected = {"1" : {"seq1" : "ACCA", "seq2" : "GATG"},
"2" : {"seq1" : "GT", "seq2" : "GA"}}
with RequiredCall(_VALIDATION_PATH, args = [msa]):
assert_equal(split_msa(msa, "112"), expected)
def test_split_msa__single_group():
msa = {"seq1" : "ACGCAT", "seq2" : "GAGTGA"}
expected = {'1' : {"seq1" : "ACGCAT", "seq2" : "GAGTGA"}}
with RequiredCall(_VALIDATION_PATH, args = [msa]):
assert_equal(split_msa(msa, "111"), expected)