def test_specs__path_is_displayed_in_exception():
def _path_is_displayed_in_exception(spec, value):
assert_raises_regexp(MakefileError, _DUMMY_PATH_STR, spec, _DUMMY_PATH,
value)
yield _path_is_displayed_in_exception, IsInt(), "foo"
yield _path_is_displayed_in_exception, IsUnsignedInt(), -1
yield _path_is_displayed_in_exception, IsFloat(), "abc"
yield _path_is_displayed_in_exception, IsBoolean(), 1
yield _path_is_displayed_in_exception, IsStr(), 1
yield _path_is_displayed_in_exception, IsNone(), 1
yield _path_is_displayed_in_exception, ValueLT(0), 1
yield _path_is_displayed_in_exception, ValueLE(0), 1
yield _path_is_displayed_in_exception, ValueGE(0), -1
yield _path_is_displayed_in_exception, ValueGT(0), -1
yield _path_is_displayed_in_exception, ValueIn([1]), 2
yield _path_is_displayed_in_exception, ValuesIntersect([1]), [2]
yield _path_is_displayed_in_exception, ValuesSubsetOf([1]), [2]
yield _path_is_displayed_in_exception, ValueMissing(), True
yield _path_is_displayed_in_exception, And(IsStr), 1
yield _path_is_displayed_in_exception, Or(IsStr), 1
yield _path_is_displayed_in_exception, Xor(IsStr, IsInt), True
yield _path_is_displayed_in_exception, Not(IsInt), 1
yield _path_is_displayed_in_exception, StringIn("abc"), 1
yield _path_is_displayed_in_exception, StringsIntersect("abc"), [1]
yield _path_is_displayed_in_exception, StringsSubsetOf("abc"), [1]
yield _path_is_displayed_in_exception, StringIsUppercase(), 1
yield _path_is_displayed_in_exception, StringStartsWith("FOO"), 1
yield _path_is_displayed_in_exception, StringEndsWith("FOO"), 1
yield _path_is_displayed_in_exception, IsListOf(IsInt), "foo"
yield _path_is_displayed_in_exception, IsDictOf(IsInt, IsInt), 1
def test_or__default_set__valid_value():
spec = Or(IsInt, ValueGT(10), default=17)
assert_equal(spec.default, 17)
def test_or__default_not_set():
spec = Or(IsInt, ValueGT(10))
assert_is(spec.default, DEFAULT_NOT_SET)
def test_or__rejects_if_both_specs_fail():
spec = Or(IsStr, IsBoolean)
assert_raises(MakefileError, spec, _DUMMY_PATH, 1)
def test_or__accepts_second_test():
spec = Or(IsStr, IsBoolean)
spec(_DUMMY_PATH, False)
def test_or__accepts_first_test():
spec = Or(IsStr, IsBoolean)
spec(_DUMMY_PATH, "Foo")
IsStr, # Not used; left for backwards compatibility
"CommonName":
IsStr, # Not used; left for backwards compatibility
"Gender":
IsStr(default=REQUIRED_VALUE),
}
}
_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": {
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,
_VALID_TARGET_NAME = \
And(_alphanum_check(whitelist="._-"),
ValueGE(2, key=len, description="at least two characters long"))
_VALIDATION_OPTIONS = {
# Sequencing platform, used to tag read-groups.
"Platform":
StringIn(("CAPILLARY", "LS454", "ILLUMINA", "SOLID", "HELICOS",
"IONTORRENT", "PACBIO"),
default="ILLUMINA"),
# Offset for quality scores in FASTQ files.
"QualityOffset":
ValueIn((33, 64, "Solexa"), default=33),
# Split a lane into multiple entries, one for each (pair of) file(s)
"SplitLanesByFilenames":
Or(IsBoolean, IsListOf(IsStr), default=True),
# Format to use when compressing FASTQ files ("gz" or "bz2")
"CompressionFormat":
ValueIn(("gz", "bz2"), default="bz2"),
"AdapterRemoval": {
"Version": ValueIn(("v1.4", "v1.5+"), default="v1.5+"),
"--pcr1": IsStr,
"--pcr2": IsStr,
"--adapter1": IsStr,
"--adapter2": IsStr,
"--maxns": IsUnsignedInt,
"--minquality": IsUnsignedInt,
"--trimns": Or(IsNone, IsBoolean),
"--trimqualities": Or(IsNone, IsBoolean),
"--collapse": Or(IsNone, IsBoolean, default=True),
"--mm": Or(IsFloat, IsUnsignedInt, default=3),
import sys
import pprint
from pypeline.common.makefile import \
MakefileError, \
read_makefile, \
WithoutDefaults, \
Or, \
IsInt, \
IsFloat, \
IsStr, \
StringStartsWith
_SPECIFICATION_OF_OPTIONS = {
StringStartsWith("--") : Or(IsInt, IsFloat),
"--min-depth" : IsInt(default = 8),
"--max-depth" : IsInt(default = 100),
}
_MAKEFILE_SPECIFICATION = {
"Defaults" : _SPECIFICATION_OF_OPTIONS,
"VCF_Files" : {
IsStr : {
"Output_File" : IsStr,
"Options" : WithoutDefaults(_SPECIFICATION_OF_OPTIONS),
}
}
}
},
},
"Filter Singletons" : {
IsStr : IsListOf(IsStr),
},
},
"Genotyping" : {
"Default" : OneOf("random", "samtools", case_sensitive = False),
"Padding" : IsInt,
"Indels" : IsBoolean,
AnyOf("MPileup", "BCFTools", "Random") : {
IsStrWithPrefix("-") : CLI_PARAMETERS,
},
"VCF_Filter" : {
"Mappability" : IsStr,
"MaxReadDepth" : Or(IsUnsignedInt, IsDictOf(IsStr, IsInt)),
IsStrWithPrefix("-") : CLI_PARAMETERS,
},
},
"MSAlignment" : {
"Enabled" : IsBoolean,
"Default" : AnyOf("mafft", case_sensitive = False),
"MAFFT" : {
"Algorithm" : AnyOf("auto", "g-ins-i", case_sensitive = False), # TODO
},
},
"Phylogenetic Inference" : {
"ExcludeGroups" : IsListOf(IsStr),
"Default" : AnyOf("raxml", "raxml-light", "examl", case_sensitive = False),
"ExaML" : {
"Threads" : IsInt,
from pypeline.common.makefile import \
read_makefile, \
IsUnsignedInt, \
IsFloat, \
IsStr, \
IsBoolean, \
And, \
Or, \
ValueIn, \
IsNone, \
StringIn, \
ValueGE, \
ValuesSubsetOf, \
IsListOf
EXCLUDEBED = Or(IsListOf(IsStr), IsStr, IsNone, default=None)
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_BED_NAME = _VALID_TARGET_NAME = \
And(_alphanum_check(whitelist=".-"),
ValueGE(2, key=len, description="at least two characters long"))