def test_306_BedDataSet_array_from_query(self):
chromstart = 10
chromend = 20
query = [(0, 15, 25), (0, 5, 13)]
dataset = peaksql.BedDataSet(DATABASE_BED, seq_length=10, stride=10)
assert np.all(
dataset.array_from_query(query, chromstart, chromend) == np.array([
[True, True, True, False, False, True, True, True, True, True]
]))
def test_310_BedDataSet_random_pos_distribution(self):
dataset = peaksql.BedDataSet(DATABASE_BED,
seq_length=10,
nr_rand_pos=100_000)
# chromosomes are of equal size, so we expect equal nr of positions for each
un_cumsum = dataset.cumsum - np.roll(dataset.cumsum, shift=1)
for count in un_cumsum[1:]:
assert 0.245 <= count / 100_000 <= 0.255
def test_302_BedDataSet_stride_sequences(self):
dataset = peaksql.BedDataSet(DATABASE_BED, seq_length=10, stride=10)
all_dna = ("AAAACCCCGGGGTTTTAAACCCGGGTTTAACCGGTTACGT" +
"TTTTGGGGCCCCAAAATTTGGGCCCAAATTGGCCAATGCA" +
"ATGCGTAGCTGATCGATGCTAGCTAGCTAGCTAGCTAAAA" +
"ATGGTGAATGTGAGTAGTGATGATGAGTGTAGTGAGGGGG")
dna_strided = [all_dna[i:i + 10] for i in range(0, len(all_dna), 10)]
dna_onehot = [
peaksql.util.sequence_to_onehot(dna) for dna in dna_strided
]
for seq, label in dataset:
assert np.sum(
np.all(seq == potential_seq) for potential_seq in dna_onehot)
def test_309_BedDataSet_random_pos_sequences(self):
dataset = peaksql.BedDataSet(DATABASE_BED,
seq_length=10,
nr_rand_pos=20)
all_dna = [
"AAAACCCCGGGGTTTTAAACCCGGGTTTAACCGGTTACGT",
"TTTTGGGGCCCCAAAATTTGGGCCCAAATTGGCCAATGCA",
"ATGCGTAGCTGATCGATGCTAGCTAGCTAGCTAGCTAAAA",
"ATGGTGAATGTGAGTAGTGATGATGAGTGTAGTGAGGGGG",
]
dna_onehot = [peaksql.util.sequence_to_onehot(dna) for dna in all_dna]
for i, (seq, label) in enumerate(dataset):
found = False
for chromosome in range(4):
for idx in range(0, 30):
if np.all(seq == dna_onehot[chromosome][idx:idx + 10]):
found = True
assert found
def test_402_Integration_PyTorch_DataLoader(self):
dataset = peaksql.BedDataSet(DATABASE_BED, nr_rand_pos=100, seq_length=3)
dataloader = DataLoader(dataset, batch_size=10)
for seq, label in dataloader:
assert tuple(seq.shape) == (10, 3, 4,)
assert tuple(label.shape) == (10, 1,)
def test_401_iterable(self):
dataset = peaksql.BedDataSet(DATABASE_BED, nr_rand_pos=100, seq_length=3)
for seq, label in dataset:
assert seq.shape == (3, 4,)
assert label.shape == (1,)
def test_308_BedDataSet_random_pos_length(self):
dataset = peaksql.BedDataSet(DATABASE_BED,
seq_length=10,
nr_rand_pos=20)
assert len(dataset) == 20
def test_305_Bed_label_fraction(self):
dataset = peaksql.BedDataSet(DATABASE_BED, seq_length=10, stride=10)
dataset.ratio = 0.4
assert all(
dataset.fraction(self.positions) == [False, True, False, True])
def test_304_Bed_label_all(self):
dataset = peaksql.BedDataSet(DATABASE_BED, seq_length=10, stride=10)
assert all(dataset.all(self.positions) == [False, False, False, True])
def test_301_BedDataSet_stride_length(self):
dataset = peaksql.BedDataSet(DATABASE_BED, seq_length=10, stride=10)
assert len(dataset) == 16