def test_construct_fails(self):
"""construction fails for invalid input"""
# fails for entries all zero
data_all_zero = [
[0.0, 0.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 0.0],
]
with self.assertRaises(ValueError):
pssm = PSSM(data_all_zero, "ACTG")
# fails for numpy.nan
data_nan = [
[nan, -0.263, -1.0, -1.322],
[-2.322, -1.678, -2.322, -1.322],
[-1.485, -1.322, -1.322, -1.322],
[-1.263, -0.737, -2.322, -0.322],
]
with self.assertRaises(ValueError):
pssm = PSSM(data_nan, "ACTG")
# fails for entries all negative numbers
data = [
[-1.322, -0.263, -1.0, -1.322],
[-2.322, -1.678, -2.322, -1.322],
[-1.485, -1.322, -1.322, -1.322],
[-1.263, -0.737, -2.322, -0.322],
]
with self.assertRaises(ValueError):
pssm = PSSM(data, "ACTG")
def test_score_seq_obj(self):
"""produce correct score from seq"""
from cogent3 import DNA
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")
seq = DNA.make_seq("".join("ACTG"[i] for i in [3, 1, 2, 0, 2, 2, 3]))
scores = pssm.score_seq(seq)
assert_allclose(scores, [-4.481, -5.703, -2.966], atol=1e-3)
def test_score_str(self):
"""produce correct score from seq"""
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")
seq = "".join("ACTG"[i] for i in [3, 1, 2, 0, 2, 2, 3])
scores = pssm.score_seq(seq)
assert_allclose(scores, [-4.481, -5.703, -2.966], atol=1e-3)
with self.assertRaises(ValueError):
pssm.score_seq(seq[:3])
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_construct_succeeds(self):
"""correctly construction from freqs"""
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")
expect = array([
[-1.322, 0.263, 1.0, -1.322],
[0.0, 0.0, 0.0, 0.0],
[-2.322, 1.678, -2.322, -1.322],
[1.485, -1.322, -1.322, -1.322],
[1.263, -0.737, -2.322, -0.322],
])
assert_allclose(pssm.array, expect, atol=1e-3)
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)
def test_construct_fails(self):
"""fails if given wrong data type or no data"""
# can't use a string
data = [["A", "A"], ["A", "A"], ["A", "A"]]
with self.assertRaises(ValueError):
got = MotifCountsArray(data, "AB")
# or a float
data = [[1.1, 2.1], [0.0, 2.1], [3.0, 4.5]]
with self.assertRaises(ValueError):
got = MotifCountsArray(data, "AB")
# or be empty
with self.assertRaises(ValueError):
got = MotifCountsArray([], "AB")
with self.assertRaises(ValueError):
got = MotifCountsArray([[], []], "AB")
data = [[2, 4], [3, 5], [4, 8]]
with self.assertRaises(ValueError):
pssm = PSSM(data, "ACGT")
def test_score_indices(self):
"""produce correct score from indexed seq"""
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")
indices = [3, 1, 2, 0, 2, 2, 3]
scores = pssm.score_indexed_seq(indices)
assert_allclose(scores, [-4.481, -5.703, -2.966], atol=1e-3)
indices = [4, 1, 2, 0, 2, 2, 3]
scores = pssm.score_indexed_seq(indices)
# log2 of (0.25 * 0.05 * 0.7 * 0.05) / .25**4 = -3.158...
assert_allclose(scores, [-3.158, -5.703, -2.966], atol=1e-3)
# fails if sequence too short
with self.assertRaises(ValueError):
pssm.score_indexed_seq(indices[:3])
