def testSetAttribute(self):
expected = 5
read = protocol.ReadAlignment()
values = read.attributes.attr['key'].values
protocol.setAttribute(values, expected)
tag, value = read.attributes.attr.items()[0]
result = value.values[0].int32_value
self.assertEqual(result, expected)
def testSetAttribute(self):
expected = 5
read = protocol.ReadAlignment()
values = read.attributes.attr['key'].values
protocol.setAttribute(values, expected)
tag, value = read.attributes.attr.items()[0]
result = value.values[0].int32_value
self.assertEqual(result, expected)
def serializeAttributes(self, msg):
"""
Sets the attrbutes of a message during serialization.
"""
attributes = self.getAttributes()
for key in attributes:
protocol.setAttribute(
msg.attributes.attr[key].values, attributes[key])
return msg
def serializeAttributes(self, msg):
"""
Sets the attrbutes of a message during serialization.
"""
attributes = self.getAttributes()
for key in attributes:
protocol.setAttribute(
msg.attributes.attr[key].values, attributes[key])
return msg
def serializeMetadataAttributes(self, msg, tier=0):
"""
Sets the attrbutes of a message for metadata during serialization.
"""
attributes = self.getAttributes()
for attribute_name in attributes:
if self.validateAttribute(attribute_name, attributes, tier) is True:
values = []
for dictionary in attributes[attribute_name]:
for key in dictionary:
values.append(dictionary[key])
protocol.setAttribute(
msg.attributes.attr[attribute_name].values, values)
return msg
def convertReadAlignment(self, read, readGroupSet, readGroupId):
"""
Convert a pysam ReadAlignment to a GA4GH ReadAlignment
"""
samFile = self.getFileHandle(self._dataUrl)
# TODO fill out remaining fields
# TODO refine in tandem with code in converters module
ret = protocol.ReadAlignment()
# ret.fragmentId = 'TODO'
ret.aligned_quality.extend(read.query_qualities)
ret.aligned_sequence = read.query_sequence
if SamFlags.isFlagSet(read.flag, SamFlags.READ_UNMAPPED):
ret.ClearField("alignment")
else:
ret.alignment.CopyFrom(protocol.LinearAlignment())
ret.alignment.mapping_quality = read.mapping_quality
ret.alignment.position.CopyFrom(protocol.Position())
ret.alignment.position.reference_name = samFile.getrname(
read.reference_id)
ret.alignment.position.position = read.reference_start
ret.alignment.position.strand = protocol.POS_STRAND
if SamFlags.isFlagSet(read.flag, SamFlags.READ_REVERSE_STRAND):
ret.alignment.position.strand = protocol.NEG_STRAND
for operation, length in read.cigar:
gaCigarUnit = ret.alignment.cigar.add()
gaCigarUnit.operation = SamCigar.int2ga(operation)
gaCigarUnit.operation_length = length
gaCigarUnit.reference_sequence = "" # TODO fix this!
ret.duplicate_fragment = SamFlags.isFlagSet(read.flag,
SamFlags.DUPLICATE_READ)
ret.failed_vendor_quality_checks = SamFlags.isFlagSet(
read.flag, SamFlags.FAILED_QUALITY_CHECK)
ret.fragment_length = read.template_length
ret.fragment_name = read.query_name
for key, value in read.tags:
# Useful for inspecting the structure of read tags
# print("{key} {ktype}: {value}, {vtype}".format(
# key=key, ktype=type(key), value=value, vtype=type(value)))
protocol.setAttribute(ret.attributes.attr[key].values, value)
if SamFlags.isFlagSet(read.flag, SamFlags.MATE_UNMAPPED):
ret.next_mate_position.Clear()
else:
ret.next_mate_position.Clear()
if read.next_reference_id != -1:
ret.next_mate_position.reference_name = samFile.getrname(
read.next_reference_id)
else:
ret.next_mate_position.reference_name = ""
ret.next_mate_position.position = read.next_reference_start
ret.next_mate_position.strand = protocol.POS_STRAND
if SamFlags.isFlagSet(read.flag, SamFlags.MATE_REVERSE_STRAND):
ret.next_mate_position.strand = protocol.NEG_STRAND
if SamFlags.isFlagSet(read.flag, SamFlags.READ_PAIRED):
ret.number_reads = 2
else:
ret.number_reads = 1
ret.read_number = -1
if SamFlags.isFlagSet(read.flag, SamFlags.FIRST_IN_PAIR):
if SamFlags.isFlagSet(read.flag, SamFlags.SECOND_IN_PAIR):
ret.read_number = 2
else:
ret.read_number = 0
elif SamFlags.isFlagSet(read.flag, SamFlags.SECOND_IN_PAIR):
ret.read_number = 1
ret.improper_placement = not SamFlags.isFlagSet(
read.flag, SamFlags.READ_PROPER_PAIR)
ret.read_group_id = readGroupId
ret.secondary_alignment = SamFlags.isFlagSet(
read.flag, SamFlags.SECONDARY_ALIGNMENT)
ret.supplementary_alignment = SamFlags.isFlagSet(
read.flag, SamFlags.SUPPLEMENTARY_ALIGNMENT)
ret.id = readGroupSet.getReadAlignmentId(ret)
return ret
def assertAlignmentsEqual(self, gaAlignment, pysamAlignment,
readGroupInfo):
if pysamAlignment.query_qualities is None:
self.assertEqual(gaAlignment.aligned_quality, [])
else:
self.assertEqual(gaAlignment.aligned_quality,
list(pysamAlignment.query_qualities))
self.assertEqual(gaAlignment.aligned_sequence,
pysamAlignment.query_sequence)
if reads.SamFlags.isFlagSet(pysamAlignment.flag,
reads.SamFlags.READ_UNMAPPED):
self.assertEqual(0, gaAlignment.alignment.ByteSize())
else:
self.assertEqual(gaAlignment.alignment.mapping_quality,
pysamAlignment.mapping_quality)
self.assertEqual(
gaAlignment.alignment.position.reference_name,
readGroupInfo.samFile.getrname(pysamAlignment.reference_id))
self.assertEqual(gaAlignment.alignment.position.position,
pysamAlignment.reference_start)
# TODO test reverseStrand on position and on
# nextMatePosition once it has been implemented.
self.assertCigarEqual(gaAlignment.alignment.cigar,
pysamAlignment.cigar)
self.assertFlag(gaAlignment.duplicate_fragment, pysamAlignment,
reads.SamFlags.DUPLICATE_READ)
self.assertFlag(gaAlignment.failed_vendor_quality_checks,
pysamAlignment, reads.SamFlags.FAILED_QUALITY_CHECK)
self.assertEqual(gaAlignment.fragment_length,
pysamAlignment.template_length)
self.assertEqual(gaAlignment.fragment_name, pysamAlignment.query_name)
compoundId = datamodel.ReadAlignmentCompoundId(
self._gaObject.getCompoundId(), pysamAlignment.query_name)
self.assertEqual(gaAlignment.id, str(compoundId))
ret = protocol.ReadAlignment()
for key, value in pysamAlignment.tags:
protocol.setAttribute(ret.attributes.attr[key].values, value)
self.assertEqual(protocol.toJson(gaAlignment.attributes),
protocol.toJson(ret.attributes))
if reads.SamFlags.isFlagSet(pysamAlignment.flag,
reads.SamFlags.MATE_UNMAPPED):
self.assertEqual(0, gaAlignment.next_mate_position.ByteSize())
else:
self.assertEqual(gaAlignment.next_mate_position.position,
pysamAlignment.next_reference_start)
if pysamAlignment.next_reference_id != -1:
self.assertEqual(
gaAlignment.next_mate_position.reference_name,
readGroupInfo.samFile.getrname(
pysamAlignment.next_reference_id))
else:
self.assertEqual(gaAlignment.next_mate_position.reference_name,
"")
if gaAlignment.number_reads == 1:
self.assertFlag(False, pysamAlignment, reads.SamFlags.READ_PAIRED)
elif gaAlignment.number_reads == 2:
self.assertFlag(True, pysamAlignment, reads.SamFlags.READ_PAIRED)
else:
# we shouldn't be setting numberReads to anything else
self.assertTrue(False)
if gaAlignment.read_number is -1:
self.assertFlag(False, pysamAlignment,
reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(False, pysamAlignment,
reads.SamFlags.SECOND_IN_PAIR)
elif gaAlignment.read_number == 0:
self.assertFlag(True, pysamAlignment, reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(False, pysamAlignment,
reads.SamFlags.SECOND_IN_PAIR)
elif gaAlignment.read_number == 1:
self.assertFlag(False, pysamAlignment,
reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(True, pysamAlignment,
reads.SamFlags.SECOND_IN_PAIR)
elif gaAlignment.read_number == 2:
self.assertFlag(True, pysamAlignment, reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(True, pysamAlignment,
reads.SamFlags.SECOND_IN_PAIR)
else:
# we shouldn't be setting readNumber to anything else
self.assertTrue(False)
self.assertFlag(not gaAlignment.improper_placement, pysamAlignment,
reads.SamFlags.READ_PROPER_PAIR)
self.assertEqual(gaAlignment.read_group_id, readGroupInfo.id)
self.assertFlag(gaAlignment.secondary_alignment, pysamAlignment,
reads.SamFlags.SECONDARY_ALIGNMENT)
self.assertFlag(gaAlignment.supplementary_alignment, pysamAlignment,
reads.SamFlags.SUPPLEMENTARY_ALIGNMENT)
def convertReadAlignment(self, read, readGroupSet, readGroupId):
"""
Convert a pysam ReadAlignment to a GA4GH ReadAlignment
"""
samFile = self.getFileHandle(self._dataUrl)
# TODO fill out remaining fields
# TODO refine in tandem with code in converters module
ret = protocol.ReadAlignment()
# ret.fragmentId = 'TODO'
ret.aligned_quality.extend(read.query_qualities)
ret.aligned_sequence = read.query_sequence
if SamFlags.isFlagSet(read.flag, SamFlags.READ_UNMAPPED):
ret.ClearField("alignment")
else:
ret.alignment.CopyFrom(protocol.LinearAlignment())
ret.alignment.mapping_quality = read.mapping_quality
ret.alignment.position.CopyFrom(protocol.Position())
ret.alignment.position.reference_name = samFile.getrname(
read.reference_id)
ret.alignment.position.position = read.reference_start
ret.alignment.position.strand = protocol.POS_STRAND
if SamFlags.isFlagSet(read.flag, SamFlags.READ_REVERSE_STRAND):
ret.alignment.position.strand = protocol.NEG_STRAND
for operation, length in read.cigar:
gaCigarUnit = ret.alignment.cigar.add()
gaCigarUnit.operation = SamCigar.int2ga(operation)
gaCigarUnit.operation_length = length
gaCigarUnit.reference_sequence = "" # TODO fix this!
ret.duplicate_fragment = SamFlags.isFlagSet(
read.flag, SamFlags.DUPLICATE_READ)
ret.failed_vendor_quality_checks = SamFlags.isFlagSet(
read.flag, SamFlags.FAILED_QUALITY_CHECK)
ret.fragment_length = read.template_length
ret.fragment_name = read.query_name
for key, value in read.tags:
# Useful for inspecting the structure of read tags
# print("{key} {ktype}: {value}, {vtype}".format(
# key=key, ktype=type(key), value=value, vtype=type(value)))
protocol.setAttribute(ret.attributes.attr[key].values, value)
if SamFlags.isFlagSet(read.flag, SamFlags.MATE_UNMAPPED):
ret.next_mate_position.Clear()
else:
ret.next_mate_position.Clear()
if read.next_reference_id != -1:
ret.next_mate_position.reference_name = samFile.getrname(
read.next_reference_id)
else:
ret.next_mate_position.reference_name = ""
ret.next_mate_position.position = read.next_reference_start
ret.next_mate_position.strand = protocol.POS_STRAND
if SamFlags.isFlagSet(read.flag, SamFlags.MATE_REVERSE_STRAND):
ret.next_mate_position.strand = protocol.NEG_STRAND
if SamFlags.isFlagSet(read.flag, SamFlags.READ_PAIRED):
ret.number_reads = 2
else:
ret.number_reads = 1
ret.read_number = -1
if SamFlags.isFlagSet(read.flag, SamFlags.FIRST_IN_PAIR):
if SamFlags.isFlagSet(read.flag, SamFlags.SECOND_IN_PAIR):
ret.read_number = 2
else:
ret.read_number = 0
elif SamFlags.isFlagSet(read.flag, SamFlags.SECOND_IN_PAIR):
ret.read_number = 1
ret.improper_placement = not SamFlags.isFlagSet(
read.flag, SamFlags.READ_PROPER_PAIR)
ret.read_group_id = readGroupId
ret.secondary_alignment = SamFlags.isFlagSet(
read.flag, SamFlags.SECONDARY_ALIGNMENT)
ret.supplementary_alignment = SamFlags.isFlagSet(
read.flag, SamFlags.SUPPLEMENTARY_ALIGNMENT)
ret.id = readGroupSet.getReadAlignmentId(ret)
return ret
def assertAlignmentsEqual(self, gaAlignment, pysamAlignment,
readGroupInfo):
if pysamAlignment.query_qualities is None:
self.assertEqual(gaAlignment.aligned_quality, [])
else:
self.assertEqual(
gaAlignment.aligned_quality,
list(pysamAlignment.query_qualities))
self.assertEqual(
gaAlignment.aligned_sequence,
pysamAlignment.query_sequence)
if reads.SamFlags.isFlagSet(
pysamAlignment.flag, reads.SamFlags.READ_UNMAPPED):
self.assertEqual(0, gaAlignment.alignment.ByteSize())
else:
self.assertEqual(
gaAlignment.alignment.mapping_quality,
pysamAlignment.mapping_quality)
self.assertEqual(
gaAlignment.alignment.position.reference_name,
readGroupInfo.samFile.getrname(pysamAlignment.reference_id))
self.assertEqual(
gaAlignment.alignment.position.position,
pysamAlignment.reference_start)
# TODO test reverseStrand on position and on
# nextMatePosition once it has been implemented.
self.assertCigarEqual(
gaAlignment.alignment.cigar,
pysamAlignment.cigar)
self.assertFlag(
gaAlignment.duplicate_fragment,
pysamAlignment, reads.SamFlags.DUPLICATE_READ)
self.assertFlag(
gaAlignment.failed_vendor_quality_checks,
pysamAlignment, reads.SamFlags.FAILED_QUALITY_CHECK)
self.assertEqual(
gaAlignment.fragment_length,
pysamAlignment.template_length)
self.assertEqual(
gaAlignment.fragment_name,
pysamAlignment.query_name)
compoundId = datamodel.ReadAlignmentCompoundId(
self._gaObject.getCompoundId(),
pysamAlignment.query_name)
self.assertEqual(gaAlignment.id, str(compoundId))
ret = protocol.ReadAlignment()
for key, value in pysamAlignment.tags:
protocol.setAttribute(ret.attributes.attr[key].values, value)
self.assertEqual(
protocol.toJson(gaAlignment.attributes),
protocol.toJson(ret.attributes))
if reads.SamFlags.isFlagSet(
pysamAlignment.flag, reads.SamFlags.MATE_UNMAPPED):
self.assertEqual(0, gaAlignment.next_mate_position.ByteSize())
else:
self.assertEqual(
gaAlignment.next_mate_position.position,
pysamAlignment.next_reference_start)
if pysamAlignment.next_reference_id != -1:
self.assertEqual(
gaAlignment.next_mate_position.reference_name,
readGroupInfo.samFile.getrname(
pysamAlignment.next_reference_id))
else:
self.assertEqual(
gaAlignment.next_mate_position.reference_name, "")
if gaAlignment.number_reads == 1:
self.assertFlag(
False, pysamAlignment, reads.SamFlags.READ_PAIRED)
elif gaAlignment.number_reads == 2:
self.assertFlag(
True, pysamAlignment, reads.SamFlags.READ_PAIRED)
else:
# we shouldn't be setting numberReads to anything else
self.assertTrue(False)
if gaAlignment.read_number is -1:
self.assertFlag(
False, pysamAlignment, reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(
False, pysamAlignment, reads.SamFlags.SECOND_IN_PAIR)
elif gaAlignment.read_number == 0:
self.assertFlag(
True, pysamAlignment, reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(
False, pysamAlignment, reads.SamFlags.SECOND_IN_PAIR)
elif gaAlignment.read_number == 1:
self.assertFlag(
False, pysamAlignment, reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(
True, pysamAlignment, reads.SamFlags.SECOND_IN_PAIR)
elif gaAlignment.read_number == 2:
self.assertFlag(
True, pysamAlignment, reads.SamFlags.FIRST_IN_PAIR)
self.assertFlag(
True, pysamAlignment, reads.SamFlags.SECOND_IN_PAIR)
else:
# we shouldn't be setting readNumber to anything else
self.assertTrue(False)
self.assertFlag(
not gaAlignment.improper_placement,
pysamAlignment, reads.SamFlags.READ_PROPER_PAIR)
self.assertEqual(
gaAlignment.read_group_id,
readGroupInfo.id)
self.assertFlag(
gaAlignment.secondary_alignment,
pysamAlignment, reads.SamFlags.SECONDARY_ALIGNMENT)
self.assertFlag(
gaAlignment.supplementary_alignment,
pysamAlignment, reads.SamFlags.SUPPLEMENTARY_ALIGNMENT)