return _validate_makefiles(config, makefiles)
def _alphanum_check(whitelist):
description = "characters a-z, A-Z, 0-9%s allowed"
description %= (", and %r" % whitelist, ) if whitelist else ""
whitelist += string.ascii_letters + string.digits
return And(IsStr(), ValuesSubsetOf(whitelist, description=description))
# Valid names for prefixes
_VALID_PREFIX_NAME = \
And(_alphanum_check(whitelist="._-*"),
Not(StringIn(["Options"] + [(s + "Reads") for s in _READ_TYPES])))
# Valid paths for prefixes; avoids some problems with e.g. Bowtie2
_VALID_PREFIX_PATH = \
And(IsStr(), Not(ValuesIntersect("\\:?\"<>|() \t\n\v\f\r")),
default=REQUIRED_VALUE)
# Valid strings for targets / samples / libraries / lanes
_VALID_TARGET_NAME = \
And(_alphanum_check(whitelist="._-"),
ValueGE(2, key=len, description="at least two characters long"))
_VALID_FEATURES_DICT = {
"Coverage":
IsBoolean(default=True),
"Depths":
def _alphanum_check(whitelist, min_len=1):
description = "characters a-z, A-Z, 0-9%s allowed"
description %= (", and %r" % whitelist, ) if whitelist else ""
whitelist += string.ascii_letters + string.digits
return And(IsStr(min_len=min_len),
ValuesSubsetOf(whitelist, description=description))
# Valid names for prefixes
_VALID_PREFIX_NAME = And(
_alphanum_check(whitelist="._-*"),
Not(StringIn(["Options"] + [(s + "Reads") for s in _READ_TYPES])),
)
# Valid paths for prefixes; avoids some problems with e.g. Bowtie2
_VALID_PREFIX_PATH = And(IsStr(),
Not(ValuesIntersect('\\:?"<>|() \t\n\v\f\r')),
default=REQUIRED_VALUE)
# Valid strings for targets / samples / libraries / lanes
_VALID_TARGET_NAME = _alphanum_check(whitelist="._-", min_len=2)
_VALID_FEATURES_DICT = {
"Coverage": IsBoolean(default=True),
"Depths": IsBoolean(default=True),
"DuplicateHist": RemovedOption(),
"RawBAM": RemovedOption(),
msa[key] = fill_dict(msa.get(key, {}), defaults)
unknown_regions = set(msa) - set(mkfile["Project"]["Regions"])
if unknown_regions:
raise MakefileError("Unknown Regions of Interest in Genotyping: %s" \
% (", ".join(unknown_regions),))
# Recursive definition of sample tree
_VALIDATION_SUBSAMPLE_KEY = And(StringStartsWith("<"),
StringEndsWith(">"))
_VALIDATION_SAMPLES_KEY = And(IsStr, Not(_VALIDATION_SUBSAMPLE_KEY))
_VALIDATION_SAMPLES = {
_VALIDATION_SAMPLES_KEY: {
"GenotypingMethod": StringIn(("reference sequence",
"random sampling",
"samtools"),
default="samtools"),
"SpeciesName": IsStr, # Not used; left for backwards compatibility
"CommonName": IsStr, # Not used; left for backwards compatibility
"Sex": IsStr(),
"Gender": IsStr(),
}
}
_VALIDATION_SAMPLES[_VALIDATION_SUBSAMPLE_KEY] = _VALIDATION_SAMPLES
# Genotyping settings; note that explicit lists must not be used here, to allow
# proper inheritance of default values. Use IsListOf instead.
_VALIDATION_GENOTYPES = {
"Padding": IsUnsignedInt,
"GenotypeEntirePrefix": IsBoolean(default=False),
"MPileup": {
def test_string_in__default_set__must_meet_spec():
with pytest.raises(ValueError):
StringIn("ABCDEFGH", default="i")
def test_string_in__handles_types(value):
spec = StringIn("ABC")
with pytest.raises(MakefileError,
match="Expected value: one of 'A', 'B', or 'C'"):
spec(_DUMMY_PATH, value)
def test_string_in__default_not_set():
spec = StringIn("ABCDEFGH")
assert spec.default is DEFAULT_NOT_SET
def test_string_in__default_set__valid_value():
spec = StringIn("ABCDEFGH", default="e")
assert spec.default == "e"
def test_string_in__case_insensitive__mixed_string__string_found():
spec = StringIn(("A", "c", "B", 1, 2, 3))
spec(_DUMMY_PATH, "a")
def test_string_in__case_insensitive__value_not_set():
spec = StringIn(("Abc", "bCe", "cdE"))
with pytest.raises(MakefileError):
spec(_DUMMY_PATH, "ABce")
def test_string_in__case_insensitive__value_in_set():
spec = StringIn(("Abc", "bCe", "cdE"))
spec(_DUMMY_PATH, "Bce")
_PATH_IN_EXCEPTION_VALUES = (
(IsInt(), "foo"),
(IsUnsignedInt(), -1),
(IsFloat(), "abc"),
(IsBoolean(), 1),
(IsStr(), 1),
(IsNone(), 1),
(ValueIn([1]), 2),
(ValuesIntersect([1]), [2]),
(ValuesSubsetOf([1]), [2]),
(ValueMissing(), True),
(And(IsStr), 1),
(Or(IsStr), 1),
(Not(IsInt), 1),
(StringIn("abc"), 1),
(StringStartsWith("FOO"), 1),
(StringEndsWith("FOO"), 1),
(IsListOf(IsInt), "foo"),
(IsDictOf(IsInt, IsInt), 1),
)
@pytest.mark.parametrize("spec, value", _PATH_IN_EXCEPTION_VALUES)
def test_specs__path_is_displayed_in_exception(spec, value):
with pytest.raises(MakefileError, match=_DUMMY_PATH_STR):
spec(_DUMMY_PATH, value)
###############################################################################
###############################################################################
# proper inheritance of default values. Use IsListOf instead.
_VALIDATION_GENOTYPES = {
"Padding": IsUnsignedInt,
"GenotypeEntirePrefix": IsBoolean(default=False),
"MPileup": {
StringStartsWith("-"): Or(IsInt, IsStr, IsNone)
},
"BCFTools": {
StringStartsWith("-"): Or(IsInt, IsStr, IsNone)
},
"Random": {
"--min-distance-to-indels": IsUnsignedInt
},
"VCF_Filter": {
"MaxReadDepth":
Or(IsUnsignedInt, IsDictOf(IsStr, IsUnsignedInt), StringIn(
("auto", ))),
"--keep-ambigious-genotypes":
IsNone,
"--min-quality":
IsUnsignedInt,
"--min-allele-frequency":
RemovedOption,
"--min-mapping-quality":
IsUnsignedInt,
"--min-read-depth":
IsUnsignedInt,
"--max-read-depth":
IsUnsignedInt,
"--min-num-alt-bases":
IsUnsignedInt,
"--min-distance-to-indels":