def test_stats_motif_enrichment():
# n_motifs = 2, n_total_input = 3, n_total_control = 5
pwm1 = PositionWeightMatrix(values=[[1, 0], [0, 0], [0, 1], [1, 1]],
name='pwm1')
pwm2 = PositionWeightMatrix(values=[[1, 0], [0, 0], [0, 1], [1, 1]],
name='pwm2')
pwm3 = PositionWeightMatrix(values=[[1, 0], [0, 0], [0, 1], [1, 1]],
name='pwm3')
pwms = [pwm1, pwm2, pwm3]
motif_sites = [[[True], [True, True], []], [[], [], []], [[], [], [True]]]
motif_sites_control = [[[True], [], [True, True, True], [], [True]],
[[], [True], [], [True], []], [[], [], [], [], []]]
enrichment_results = motif_enrichment(
pwms=pwms,
motif_sites=motif_sites,
motif_sites_control=motif_sites_control)
assert len(enrichment_results) == 3
assert enrichment_results[0].n_input == 2
assert enrichment_results[0].n_control == 3
assert enrichment_results[0].fold_change == pytest.approx(2 * 5 / 3 / 3)
assert enrichment_results[1].n_input == 0
assert enrichment_results[1].n_control == 2
assert enrichment_results[1].fold_change == 0
assert enrichment_results[2].n_input == 1
assert enrichment_results[2].n_control == 0
def test_pwms_init(motif_root):
with pytest.raises(ValueError):
MotifPwms(pwms=[PositionWeightMatrix([[1], [2], [3], [4]]), 's', True])
pwm = PositionWeightMatrix([[1], [2], [3], [4]])
name = 'motif_set'
pwms = MotifPwms(pwms=[pwm], name=name)
assert len(pwms) == 1
assert pwms.name == name
pwms = MotifPwms(name=name)
pwms.read_motifscan_pwms(
os.path.join(motif_root, 'test', 'test_pwms.motifscan'))
assert len(pwms) == 2
with pytest.raises(MotifSetNotFoundError):
load_built_pwms('uninstalled_motif_set', 'hg19')
def test_pwm_raw_score():
pwm = PositionWeightMatrix([[1.35, 0.21, -5.23], [0.07, -0.21, 0.6],
[2.15, 2.22, -0.84], [-2.64, -1.89, 5.47]])
assert pwm.max_raw_score == pytest.approx(9.84)
assert pwm.max_raw_score == pytest.approx(9.84)
assert pwm.min_raw_score == pytest.approx(-9.76)
assert pwm.min_raw_score == pytest.approx(-9.76)
def test_pwm_score():
pwm = PositionWeightMatrix([[1.35, 0.21, -5.23], [0.07, -0.21, 0.6],
[2.15, 2.22, -0.84], [-2.64, -1.89, 5.47]])
with pytest.raises(ValueError):
pwm.score("")
with pytest.raises(ValueError):
pwm.score("NNNN")
assert pwm.score("NNN") == 0
assert pwm.score("AGT") == pytest.approx(0.918699)
assert pwm.score("ANT") == pytest.approx(0.693089)
assert pwm.score("CTA") == pytest.approx(-0.716463)
def test_pwms_write_motifscan_pwms(tmp_dir):
pwm = PositionWeightMatrix([[1], [2], [3], [4]],
name='motif1',
matrix_id='id1',
cutoffs={
'1e-2': 0.1,
'1e-3': 0.4
})
name = 'motif_set'
pwms = MotifPwms(pwms=[pwm], name=name)
pwms_path = os.path.join(tmp_dir, 'test_pwm.motifscan')
pwms.write_motifscan_pwms(pwms_path)
assert os.path.isfile(pwms_path)
def test_set_cutoffs():
pwm = PositionWeightMatrix([[1], [2], [3], [4]])
name = 'motif_set'
pwms = MotifPwms(pwms=[pwm], name=name)
pwms.set_cutoffs([range(1, 1000001)])
assert len(pwm.cutoffs) == 5
assert pwm.cutoffs['1e-2'] == 990000
assert pwm.cutoffs['1e-3'] == 999000
assert pwm.cutoffs['1e-4'] == 999900
assert pwm.cutoffs['1e-5'] == 999990
assert pwm.cutoffs['1e-6'] == 999999
with pytest.raises(ValueError):
pwms.set_cutoffs([[1], [2]])
with pytest.raises(ValueError):
pwms.set_cutoffs([1, 2, 3])
with pytest.raises(ValueError):
pwms.set_cutoffs([[1, 2, 3]])
def test_scan_motif(genome_root):
genome_path = os.path.join(genome_root, 'test')
genome = Genome(name='test', path=genome_path)
regions = [GenomicRegion(chrom='chr1', start=2, end=5)]
pwms = MotifPwms()
pwm = PositionWeightMatrix([[1, 0], [0, 1], [0, 0], [1, 0]],
cutoffs={
'1e-3': 0.5,
'1e-4': 1
})
pwms.append(pwm)
scanner = Scanner(genome=genome,
regions=regions,
window_size=4,
p_value='1e-4')
sites = scanner.scan_motifs(pwms)
print(sites)
assert len(sites[0]) == 1
assert len(sites[0][0]) == 1
scanner = Scanner(genome=genome,
regions=regions,
window_size=4,
p_value='1e-3')
sites = scanner.scan_motifs(pwms)
print(sites)
assert len(sites[0][0]) == 3
scanner = Scanner(genome=genome,
regions=regions,
window_size=4,
p_value='1e-2')
with pytest.raises(ValueError):
scanner.scan_motifs(pwms)
scanner = Scanner(genome=genome,
regions=regions,
window_size=4,
p_value='1e-3',
remove_dup=False)
sites = scanner.scan_motifs(pwms)
assert len(sites[0][0]) == 5
def read_motifscan_pwms(self, path):
"""Read PWMs in MotifScan format.
MotifScan PWM Example:
>MA0006.1 Ahr::Arnt PWM
A [-0.85815 -5.68647 -5.68647 -5.68647 -5.68647 -5.68647]
C [ 0.48657 -5.32257 1.53966 -5.32257 -5.32257 -5.32257]
G [-0.90016 1.53922 -5.32301 1.53922 -5.32301 1.58174]
T [ 0.43981 -1.93828 -1.93828 -1.93828 1.21696 -5.68779]
Cutoff_p1e-3 0.55403
Cutoff_p1e-4 0.82985
Cutoff_p1e-5 1.0
Parameters
----------
path : str
The file path to read the MotifScan PWMs.
Raises
------
PwmsMotifScanFormatError
If the file does not strictly follow the MotifScan PWMs format.
"""
logger.debug(f"Reading MotifScan PWMs from {path}")
header_pattern = re.compile(r"^>(\S+)\t(\S+)\tPWM$")
matrix_pattern = re.compile(r"^([ACGT]) \[(.+)\]$")
cutoff_pattern = re.compile(r"^Cutoff_p(\S+)\t(\S+)")
pwms = []
line_num = 0
# expect_flag: 0=header, 1=matrix, 2=cutoff, 3=cutoff or header
# 1 header line + 4 matrix line + at least 1 cutoff line
expect_flag = 0
with open(path, 'r') as fin:
for line in fin:
line_num += 1
line = line.strip()
if not line: # skip blank lines
continue
m_header = header_pattern.match(line)
m_matrix = matrix_pattern.match(line)
m_cutoff = cutoff_pattern.match(line)
# format checker
if m_header:
if expect_flag != 0 and expect_flag != 3:
raise PwmsMotifScanFormatError(line_num, line)
elif m_matrix:
if expect_flag != 1:
raise PwmsMotifScanFormatError(line_num, line)
elif m_cutoff:
if expect_flag != 2 and expect_flag != 3:
raise PwmsMotifScanFormatError(line_num, line)
else: # does not match any pattern
raise PwmsMotifScanFormatError(line_num, line)
if m_header:
if expect_flag == 3: # already got a pwm and save it
pwm = PositionWeightMatrix(values=values,
name=name,
matrix_id=matrix_id,
cutoffs=cutoffs)
pwms.append(pwm)
matrix_id = m_header.group(1)
name = m_header.group(2)
n_matrix = 0
values = []
cutoffs = {}
expect_flag = 1 # found header, expect matrix line next
elif m_matrix:
base = m_matrix.group(1)
if base != bases[n_matrix]:
raise PwmsMotifScanFormatError(line_num, line)
tmp_values = m_matrix.group(2).split()
try:
values.append(list(map(float, tmp_values)))
except (ValueError, TypeError):
raise PwmsMotifScanFormatError(line_num, line)
n_matrix += 1
if n_matrix == 4:
# got 4 matrix line, expect cutoff line next
expect_flag = 2
elif m_cutoff:
p = m_cutoff.group(1)
cutoff = m_cutoff.group(2)
cutoffs[p] = float(cutoff)
# got first cutoff, expect cutoff or header next
if expect_flag == 2:
expect_flag = 3
# check whether the last matrix is complete
if expect_flag == 1 or expect_flag == 2:
raise PwmsMotifScanFormatError(line_num + 1, '')
if expect_flag == 3:
pwm = PositionWeightMatrix(values=values,
name=name,
matrix_id=matrix_id,
cutoffs=cutoffs)
pwms.append(pwm)
self.extend(pwms)
logger.debug(f"Found {len(pwms)} MotifScan PWMs")