def test_bedrecord__freeze__add_new_value():
record_str = "my_contig\t12\t345\tmy_name"
record = BEDRecord(record_str)
frozen_record = record.freeze()
with pytest.raises(AttributeError):
frozen_record.score = 1
assert str(frozen_record) == record_str
def test_bedrecord__freeze__modify_existing_value():
record_str = "my_contig\t12\t345\tmy_name"
record = BEDRecord(record_str)
frozen_record = record.freeze()
with pytest.raises(TypeError):
frozen_record.start = 1
assert str(frozen_record) == record_str
def test_bedrecord__setters__3_fields():
record = BEDRecord("my_contig\t12\t345")
record.contig = "chrZ"
assert_equal(record.contig, "chrZ")
record.end += 20
assert_equal(record.end, 365)
assert_equal(str(record), "chrZ\t12\t365")
assert_equal(repr(record), "BEDRecord(contig='chrZ', start=12, end=365)")
def test_bedrecord__setters__3_fields():
record = BEDRecord("my_contig\t12\t345")
record.contig = "chrZ"
assert record.contig == "chrZ"
record.end += 20
assert record.end == 365
assert str(record) == "chrZ\t12\t365"
assert repr(record) == "BEDRecord(contig='chrZ', start=12, end=365)"
def test_bedrecord__cmp():
record_1_txt = "my_contig\t12\t345\tmy_name\t-3\t-\tfoo"
record_1 = BEDRecord(record_1_txt)
record_2 = BEDRecord("chrZ\t132\t4345\tchrZ_region\t0\t+\tbar")
for idx in xrange(len(record_2)):
record_tmp = BEDRecord(record_1_txt)
assert_equal(record_1, record_tmp)
record_tmp[idx] = record_2[idx]
assert_not_equal(record_1, record_tmp)
record_tmp[idx] = record_1[idx]
assert_equal(record_1, record_tmp)
def test_bedrecord__setters__3_fields():
record = BEDRecord("my_contig\t12\t345")
record.contig = "chrZ"
assert_equal(record.contig, "chrZ")
record.end += 20
assert_equal(record.end, 365)
assert_equal(str(record), "chrZ\t12\t365")
assert_equal(repr(record),
"BEDRecord(contig='chrZ', start=12, end=365)")
def _stat_areas_of_interest(cls, prefixes):
"""Returns (size, number of named intervals, total number of intervals)
for a set of areas of interest."""
areas_of_interest = {}
for (prefix_name, prefix) in prefixes.iteritems():
prefix_label = prefix.get("Label", prefix_name)
for (roi_name, roi_filename) in prefix.get("RegionsOfInterest",
{}).iteritems():
count, names, size = 0, set(), 0
with open(roi_filename) as handle:
for line in handle:
bed = BEDRecord(line)
names.add(bed.name if len(bed) >= 4 else (bed.contig +
"*"))
size += (bed.end - bed.start)
count += 1
areas_of_interest[(prefix_name, roi_name)] = {
"Size": size,
"NFeatures": len(names),
"NIntervals": count,
"Genome": prefix["Name"],
"Name": roi_name,
"Label": "%s:%s" % (prefix_label, roi_name),
"Path": roi_filename
}
return areas_of_interest
def test_bedrecord__constructor__3_fields():
text = "my_contig\t12\t345"
record = BEDRecord(text)
assert len(record) == 3
assert str(record) == text
assert repr(record) == "BEDRecord(contig='my_contig', start=12, end=345)"
def test_bedrecord__setters__unset_fields__at_end():
record = BEDRecord("my_contig\t12\t345")
record.name = "my_region"
assert_equal(record.name, "my_region")
record.score = -13
assert_equal(record.score, -13)
record.strand = '-'
assert_equal(record.strand, '-')
assert_equal(str(record), "my_contig\t12\t345\tmy_region\t-13\t-")
assert_equal(
repr(record), "BEDRecord(contig='my_contig', start=12, end=345, "
"name='my_region', score=-13, strand='-')")
def test_bedrecord__setters__unset_fields__at_end():
record = BEDRecord("my_contig\t12\t345")
record.name = "my_region"
assert_equal(record.name, "my_region")
record.score = -13
assert_equal(record.score, -13)
record.strand = '-'
assert_equal(record.strand, '-')
assert_equal(str(record), "my_contig\t12\t345\tmy_region\t-13\t-")
assert_equal(repr(record),
"BEDRecord(contig='my_contig', start=12, end=345, "
"name='my_region', score=-13, strand='-')")
def test_bedrecord__copy():
record_1_txt = "my_contig\t12\t345\tmy_name\t-3\t-"
record_1 = BEDRecord(record_1_txt)
record_2 = copy.copy(record_1)
record_2.name = "my_clone"
assert str(record_1) == record_1_txt
assert str(record_2) == "my_contig\t12\t345\tmy_clone\t-3\t-"
def test_bedrecord__setters__unset_fields__at_end():
record = BEDRecord("my_contig\t12\t345")
record.name = "my_region"
assert record.name == "my_region"
record.score = -13
assert record.score == -13
record.strand = "-"
assert record.strand == "-"
assert str(record) == "my_contig\t12\t345\tmy_region\t-13\t-"
assert (
repr(record) == "BEDRecord(contig='my_contig', start=12, end=345, "
"name='my_region', score=-13, strand='-')"
)
def test_bedrecord__copy():
record_1_txt = "my_contig\t12\t345\tmy_name\t-3\t-"
record_1 = BEDRecord(record_1_txt)
record_2 = copy.copy(record_1)
record_2.name = "my_clone"
assert_equal(str(record_1), record_1_txt)
assert_equal(str(record_2), "my_contig\t12\t345\tmy_clone\t-3\t-")
def test_bedrecord__constructor__3_fields():
text = "my_contig\t12\t345"
record = BEDRecord(text)
assert_equal(len(record), 3)
assert_equal(str(record), text)
assert_equal(repr(record),
"BEDRecord(contig='my_contig', start=12, end=345)")
def parse_intervals(genotype):
records = {}
for line in genotype.header:
match = _VCF_DICT.match(line)
if not match:
continue
key, values = match.groups()
if key == "contig":
values = dict(pair.split("=", 1) for pair in values.split(","))
record = BEDRecord()
record.contig = values["ID"]
record.start = 0
record.end = int(values["length"])
record.name = record.contig
record.strand = "+"
records[record.contig] = [record]
if not records:
sys.stderr.write("ERROR: List of contigs not found in VCF header; "
"specifying --intervals is required!\n")
return None
return records
def test_bedrecord__accessors__6_fields__getitem():
record = BEDRecord("my_contig\t12\t345\tmy_name\t-3\t-")
assert record[0] == "my_contig"
assert record[1] == 12
assert record[2] == 345
assert record[3] == "my_name"
assert record[4] == -3
assert record[5] == "-"
def test_bedrecord__accessors__6_fields():
record = BEDRecord("my_contig\t12\t345\tmy_name\t-3\t-")
assert record.contig == "my_contig"
assert record.start == 12
assert record.end == 345
assert record.name == "my_name"
assert record.score == -3
assert record.strand == "-"
def test_bedrecord__accessors__6_fields__getitem():
record = BEDRecord("my_contig\t12\t345\tmy_name\t-3\t-")
assert_equal(record[0], "my_contig")
assert_equal(record[1], 12)
assert_equal(record[2], 345)
assert_equal(record[3], "my_name")
assert_equal(record[4], -3)
assert_equal(record[5], "-")
def test_bedrecord__accessors__6_fields():
record = BEDRecord("my_contig\t12\t345\tmy_name\t-3\t-")
assert_equal(record.contig, "my_contig")
assert_equal(record.start, 12)
assert_equal(record.end, 345)
assert_equal(record.name, "my_name")
assert_equal(record.score, -3)
assert_equal(record.strand, "-")
def test_bedrecord__deepcopy():
record_1_txt = "my_contig\t12\t345\tmy_name\t-3\t-"
record_1 = BEDRecord(record_1_txt)
record_1[6] = ["foo"]
record_2 = copy.deepcopy(record_1)
record_2[6][0] = "bar"
assert_equal(str(record_1), record_1_txt + "\t['foo']")
assert_equal(str(record_2), record_1_txt + "\t['bar']")
def test_bedrecord__constructor__6_fields():
text = "my_contig\t12\t345\tmy_name\t-3\t-"
record = BEDRecord(text)
assert_equal(len(record), 6)
assert_equal(str(record), text)
assert_equal(
repr(record), "BEDRecord(contig='my_contig', start=12, "
"end=345, name='my_name', score=-3, strand='-')")
def test_bedrecord__setters__type_errors():
record = BEDRecord("my_contig\t12\t345\tname\t0\t+")
assert_raises(ValueError, lambda: setattr(record, "contig", 17))
assert_raises(ValueError, lambda: setattr(record, "start", "foo"))
assert_raises(ValueError, lambda: setattr(record, "end", "foo"))
assert_raises(ValueError, lambda: setattr(record, "name", 17.3))
assert_raises(ValueError, lambda: setattr(record, "score", "foo"))
assert_raises(ValueError, lambda: setattr(record, "strand", "foo"))
def test_bedrecord__accessors__3_fields():
record = BEDRecord("my_contig\t12\t345")
assert_equal(record.contig, "my_contig")
assert_equal(record.start, 12)
assert_equal(record.end, 345)
assert_raises(IndexError, lambda: record.name)
assert_raises(IndexError, lambda: record.score)
assert_raises(IndexError, lambda: record.strand)
def test_bedrecord__accessors__3_fields():
record = BEDRecord("my_contig\t12\t345")
assert record.contig == "my_contig"
assert record.start == 12
assert record.end == 345
with pytest.raises(IndexError):
record.name()
with pytest.raises(IndexError):
record.score()
with pytest.raises(IndexError):
record.strand()
def test_bedrecord__constructor__extra_fields():
text = "my_contig\t12\t345\tmy_name\t-3\t-\tfoo\tbar"
record = BEDRecord(text)
assert len(record) == 8
assert str(record) == text
assert (
repr(record) == "BEDRecord(contig='my_contig', start=12, "
"end=345, name='my_name', score=-3, strand='-', "
"'foo', 'bar')"
)
def parse_intervals(genotype):
records = {}
for line in genotype.header:
match = _VCF_DICT.match(line)
if not match:
continue
key, values = match.groups()
if key == "contig":
values = dict(pair.split("=", 1) for pair in values.split(","))
record = BEDRecord()
record.contig = values["ID"]
record.start = 0
record.end = int(values["length"])
record.name = record.contig
record.strand = "+"
records[record.contig] = [record]
if not records:
sys.stderr.write("ERROR: List of contigs not found in VCF header; "
"specifying --intervals is required!\n")
return None
return records
def test_bedrecord__setters__type_errors():
record = BEDRecord("my_contig\t12\t345\tname\t0\t+")
with pytest.raises(ValueError):
record.contig = 17
with pytest.raises(ValueError):
record.start = "foo"
with pytest.raises(ValueError):
record.end = "foo"
with pytest.raises(ValueError):
record.name = 17.3
with pytest.raises(ValueError):
record.score = "foo"
with pytest.raises(ValueError):
record.strand = "foo"
def test_bedrecord__constructor__invalid_values():
tmpl = "\t1\t%s\t\t0\t-"
BEDRecord(tmpl % (1,)) # check template
with pytest.raises(BEDError):
BEDRecord(tmpl % ("not a number",))
def test_bedrecord__setters__unset_fields__after_end():
record = BEDRecord("")
record.strand = "-"
assert_equal(str(record), "\t0\t0\t\t0\t-")
record = BEDRecord("my_name")
record.strand = "-"
assert_equal(str(record), "my_name\t0\t0\t\t0\t-")
record = BEDRecord("my_name\t17")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t0\t\t0\t-")
record = BEDRecord("my_name\t17\t258")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\t\t0\t-")
record = BEDRecord("my_name\t17\t258\tregion")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\tregion\t0\t-")
record = BEDRecord("my_name\t17\t258\tregion\t33")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\tregion\t33\t-")
record = BEDRecord("my_name\t17\t258\tregion\t33\t+")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\tregion\t33\t-")
def test_bedrecord__constructor__empty_string():
record = BEDRecord("")
assert_equal(len(record), 0)
assert_equal(str(record), "")
assert_equal(repr(record), "BEDRecord()")
def test_bedrecord__constructor__defaults():
record = BEDRecord()
assert_equal(len(record), 0)
assert_equal(str(record), "")
assert_equal(repr(record), "BEDRecord()")
def test_bedrecord__accessors__extra_fields():
text = "my_contig\t12\t345\tmy_name\t-3\t-\tfoo\tbar"
record = BEDRecord(text)
assert_equal(record[6], "foo")
assert_equal(record[7], "bar")
def test_bedrecord__constructor__empty_string():
record = BEDRecord("")
assert len(record) == 0
assert str(record) == ""
assert repr(record) == "BEDRecord()"
def test_bedrecord__setters__unset_fields__after_end():
record = BEDRecord("")
record.strand = "-"
assert_equal(str(record), "\t0\t0\t\t0\t-")
record = BEDRecord("my_name")
record.strand = "-"
assert_equal(str(record), "my_name\t0\t0\t\t0\t-")
record = BEDRecord("my_name\t17")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t0\t\t0\t-")
record = BEDRecord("my_name\t17\t258")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\t\t0\t-")
record = BEDRecord("my_name\t17\t258\tregion")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\tregion\t0\t-")
record = BEDRecord("my_name\t17\t258\tregion\t33")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\tregion\t33\t-")
record = BEDRecord("my_name\t17\t258\tregion\t33\t+")
record.strand = "-"
assert_equal(str(record), "my_name\t17\t258\tregion\t33\t-")
