def test_write_tabular_distance_matrix(self):
"""correctly writes tabular data for DistanceMatrix"""
data = {(0, 0): 0, (0, 1): 4, (1, 0): 4, (1, 1): 0}
matrix = DistanceMatrix(data)
loader = io_app.load_tabular(sep="\t")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(matrix, identifier=outpath)
new = loader(outpath)
# when written to file in tabular form
# the loaded table will have dim-1 dim-2 as column labels
# and the key-values pairs listed below; in dict form...
expected = {
0: {
"dim-1": 0,
"dim-2": 1,
"value": 4
},
1: {
"dim-1": 1,
"dim-2": 0,
"value": 4
},
}
self.assertEqual(expected, new.to_dict())
def test_write_tabular_pssm(self):
"""correctly writes tabular data for PSSM"""
# data from test_profile
data = numpy.array([
[0.1, 0.3, 0.5, 0.1],
[0.25, 0.25, 0.25, 0.25],
[0.05, 0.8, 0.05, 0.1],
[0.7, 0.1, 0.1, 0.1],
[0.6, 0.15, 0.05, 0.2],
])
pssm = PSSM(data, "ACTG")
loader = io_app.load_tabular(sep="\t")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(pssm, identifier=outpath)
new = loader(outpath)
expected = safe_log(data) - safe_log(
numpy.array([0.25, 0.25, 0.25, 0.25]))
for i in range(len(expected)):
j = i // 4
self.assertTrue(
numpy.isclose(new.array[i][2],
expected[j][i - j],
atol=0.0001))
def test_composes_with_write_tabular(self):
"""correctly links to tabular"""
with TemporaryDirectory(dir=".") as dirname:
writer = io.write_tabular(dirname)
dist_calc = dist_app.fast_slow_dist(distance="hamming",
moltype="protein")
_ = dist_calc + writer
def test_load_tabular_table(self):
"""correctly loads tabular data"""
rows = [[1, 2], [3, 4], [5, 6.5]]
table = Table(["A", "B"], data=rows)
loader = io_app.load_tabular(sep="\t", as_type="table")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(table, identifier=outpath)
new = loader(outpath)
self.assertEqual(table.to_dict(), new.to_dict())
def test_load_tabular_distance_matrix(self):
"""correctly loads tabular data for DistanceMatrix"""
data = {(0, 0): 0, (0, 1): 4, (1, 0): 4, (1, 1): 0}
matrix = DistanceMatrix(data)
loader = io_app.load_tabular(sep="\t", as_type="distances")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(matrix, identifier=outpath)
new = loader(outpath)
self.assertEqual(matrix.to_dict(), new.to_dict())
def test_load_tabular_motif_freqs_array(self):
"""correctly loads tabular data for MotifFreqsArray"""
data = [[0.3333, 0.6667], [0.3750, 0.625], [0.3333, 0.6667]]
mfa = MotifFreqsArray(data, "AB")
loader = io_app.load_tabular(sep="\t", as_type="motif_freqs")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(mfa, identifier=outpath)
new = loader(outpath)
self.assertEqual(mfa.to_dict(), new.to_dict())
def test_load_tabular_motif_counts_array(self):
"""correctly loads tabular data for MotifCountsArray"""
data = [[2, 4], [3, 5], [4, 8]]
mca = MotifCountsArray(data, "AB")
loader = io_app.load_tabular(sep="\t", as_type="motif_counts")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(mca, identifier=outpath)
new = loader(outpath)
self.assertEqual(mca.to_dict(), new.to_dict())
def test_functions_as_composable(self):
"""works as a composable app"""
from pathlib import Path
loader = io.load_aligned(moltype="dna", format="paml")
dist = dist_app.fast_slow_dist("hamming", moltype="dna")
with TemporaryDirectory(dir=".") as dirname:
dirname = Path(dirname)
writer = io.write_tabular(dirname)
proc = loader + dist + writer
_ = proc("data/brca1_5.250.paml")
output = dirname / "brca1_5.250.tsv"
self.assertTrue(output.exists())
def test_write_tabular_motif_freqs_array(self):
"""correctly writes tabular data for MotifFreqsArray"""
data = [[0.3333, 0.6667], [0.3750, 0.625], [0.3333, 0.6667]]
mfa = MotifFreqsArray(data, "AB")
loader = io_app.load_tabular(sep="\t")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(mfa, identifier=outpath)
new = loader(outpath)
# when written to file in tabular form
# the loaded table will have dim-1 dim-2 as column labels
# and the key-values pairs listed below; in dict form...
expected = {
0: {
"dim-1": 0,
"dim-2": "A",
"value": 0.3333
},
1: {
"dim-1": 0,
"dim-2": "B",
"value": 0.6667
},
2: {
"dim-1": 1,
"dim-2": "A",
"value": 0.3750
},
3: {
"dim-1": 1,
"dim-2": "B",
"value": 0.6250
},
4: {
"dim-1": 2,
"dim-2": "A",
"value": 0.3333
},
5: {
"dim-1": 2,
"dim-2": "B",
"value": 0.6667
},
}
self.assertEqual(expected, new.to_dict())
def test_load_tabular_pssm(self):
"""correctly loads tabular data for PSSM"""
# data from test_profile
data = [
[0.1, 0.3, 0.5, 0.1],
[0.25, 0.25, 0.25, 0.25],
[0.05, 0.8, 0.05, 0.1],
[0.7, 0.1, 0.1, 0.1],
[0.6, 0.15, 0.05, 0.2],
]
pssm = PSSM(data, "ACTG")
loader = io_app.load_tabular(sep="\t", as_type="pssm")
with TemporaryDirectory(dir=".") as dirname:
writer = io_app.write_tabular(data_path=dirname, format="tsv")
outpath = join(dirname, "delme.tsv")
writer.write(pssm, identifier=outpath)
new = loader(outpath)
assert_allclose(pssm.array, new.array, atol=0.0001)
