def test_identity(self):
i1, i2, i3, i4 = P.closed(0, 1), P.openclosed(0, 1), P.closedopen(0, 1), P.open(0, 1)
assert P.from_string(P.to_string(i1), int) == i1
assert P.from_string(P.to_string(i2), int) == i2
assert P.from_string(P.to_string(i3), int) == i3
assert P.from_string(P.to_string(i4), int) == i4
def get_compatible_isoforms_stitcher(mol_list, isoform_dict_json,
refskip_dict_json, h):
isoform_dict = P.IntervalDict()
for i, s in isoform_dict_json.items():
isoform_dict[P.from_string(i, conv=int)] = set(s.split(','))
refskip_dict = P.IntervalDict()
for i, s in refskip_dict_json.items():
refskip_dict[P.from_string(i, conv=int)] = set(s.split(','))
compatible_isoforms_trie = dict()
new_mol_list = []
for success, m in mol_list:
if not success:
if type(m) is str:
new_mol_list.append((success, m))
else:
new_mol_list.append((success, m.to_string()))
continue
mol = pysam.AlignedRead.fromstring(m, h)
i = interval(intervals_extract(mol.get_reference_positions()))
refskip_cigar = [
t[0] for t in mol.cigartuples if t[1] > 0 and t[0] in [2, 3]
]
blocks = mol.get_blocks()
j = []
for n in range(len(blocks) - 1):
if refskip_cigar[n] == 3:
j.append((blocks[n][1], blocks[n + 1][0]))
j = interval(j)
set_list = [
s for k, s in isoform_dict.get(i, default={'intronic'}).items()
if len(list(P.iterate(k, step=1))) > 4
]
set_refskip_list = [
s for k, s in refskip_dict.get(j, default={'intronic'}).items()
if len(list(P.iterate(k, step=1))) > 4
]
if {'intronic'} in set_list:
if len(set_list) > 1:
del set_list[set_list.index({'intronic'})]
if {'intronic'} in set_refskip_list:
if len(set_refskip_list) > 1:
del set_refskip_list[set_refskip_list.index({'intronic'})]
try:
if len(set_refskip_list) > 0:
mol.set_tag(
'CT', ','.join(
list(
set.intersection(*set_list).intersection(
*set_refskip_list))))
else:
mol.set_tag('CT', ','.join(list(set.intersection(*set_list))))
new_mol_list.append((success, mol.to_string()))
except:
continue
return new_mol_list
def test_parameters(self):
i1, i2, i3, i4 = '<"0"-"1">', '<!"0"-"1">', '<"0"-"1"!>', '<!"0"-"1"!>'
params = {
'conv': lambda s: int(s[1:-1]),
'disj': ' or ',
'sep': '-',
'left_open': '<!',
'left_closed': '<',
'right_open': '!>',
'right_closed': '>',
'pinf': r'\+oo',
'ninf': '-oo',
}
assert P.from_string(i1, **params) == P.closed(0, 1)
assert P.from_string(i2, **params) == P.openclosed(0, 1)
assert P.from_string(i3, **params) == P.closedopen(0, 1)
assert P.from_string(i4, **params) == P.open(0, 1)
assert P.from_string('<!!>', **params) == P.empty()
assert P.from_string('<"1">', **params) == P.singleton(1)
assert P.from_string('<!-oo-"1">', **params) == P.openclosed(-P.inf, 1)
assert P.from_string('<"1"-+oo!>', **params) == P.closedopen(1, P.inf)
assert P.from_string('<"0"-"1"> or <"2"-"3">', **params) == P.closed(0, 1) | P.closed(2, 3)
def get_compatible_isoforms_stitcher(mol_list, isoform_dict_json, h):
isoform_dict = P.IntervalDict()
for i, s in isoform_dict_json.items():
isoform_dict[P.from_string(i, conv=int)] = set(s.split(','))
compatible_isoforms_trie = dict()
new_mol_list = []
for success, mol in [(s, pysam.AlignedRead.fromstring(m, h))
for s, m in mol_list]:
if not success:
if type(mol) == 'str':
new_mol_list.append((success, mol))
else:
new_mol_list.append((success, mol.to_string()))
continue
i = interval(intervals_extract(mol.get_reference_positions()))
set_list = [
s for k, s in isoform_dict.get(i, default={'intronic'}).items()
if len(list(P.iterate(k, step=1))) > 4
]
if {'intronic'} in set_list:
if len(set_list) > 1:
del set_list[set_list.index({'intronic'})]
try:
mol.set_tag('CT', ','.join(list(set.intersection(*set_list))))
new_mol_list.append((success, mol.to_string()))
except:
continue
return new_mol_list
def test_conv_is_required(self):
with pytest.raises(Exception):
P.from_string('[1,2]', None)
def test_unions(self):
assert P.from_string('[0,1] | [2,3]', int) == P.closed(0, 1) | P.closed(2, 3)
def test_infinities(self):
assert P.from_string('(-inf,1]', int) == P.openclosed(-P.inf, 1)
assert P.from_string('[1,+inf)', int) == P.closedopen(1, P.inf)
def test_empty(self):
assert P.from_string('()', int) == P.empty()
def test_singleton(self):
assert P.from_string('[0]', int) == P.singleton(0)
def test_bounds(self):
assert P.from_string('[0,1]', int) == P.closed(0, 1)
assert P.from_string('(0,1]', int) == P.openclosed(0, 1)
assert P.from_string('[0,1)', int) == P.closedopen(0, 1)
assert P.from_string('(0,1)', int) == P.open(0, 1)
def time_interval_from_str(bytes_string):
converter = lambda s: datetime.strptime(s, time_format).time()
return P.from_string(bytes_string.decode("utf-8"), conv=converter)