def setUp(self):
self.bam = test_utils.genomics_core_testdata('test.bam')
self.options = reads_pb2.SamReaderOptions()
def setUp(self):
self.bam = test_utils.genomics_core_testdata('test.bam')
self.options = reads_pb2.SamReaderOptions()
self.indexed_options = reads_pb2.SamReaderOptions(
index_mode=index_pb2.INDEX_BASED_ON_FILENAME)
def __init__(self,
input_path,
ref_path=None,
read_requirements=None,
parse_aux_fields=False,
hts_block_size=None,
downsample_fraction=None,
random_seed=None):
"""Initializes a NativeSamReader.
Args:
input_path: str. A path to a resource containing SAM/BAM/CRAM records.
Currently supports SAM text format, BAM binary format, and CRAM.
ref_path: optional str or None. Only used for CRAM decoding, and only
necessary if the UR encoded path in the CRAM itself needs to be
overridden. If provided, we will tell the CRAM decoder to use this FASTA
for the reference sequence.
read_requirements: optional ReadRequirement proto. If not None, this proto
is used to control which reads are filtered out by the reader before
they are passed to the client.
parse_aux_fields: optional bool, defaulting to False. If False, we do not
parse the auxiliary fields of the SAM/BAM/CRAM records (see SAM spec for
details). Parsing the aux fields is unnecessary for many applications,
and adds a significant parsing cost to access. If you need these aux
fields, set parse_aux_fields to True and these fields will be parsed and
populate the appropriate Read proto fields (e.g., read.info).
hts_block_size: int or None. If specified, this configures the block size
of the underlying htslib file object. Larger values (e.g. 1M) may be
beneficial for reading remote files. If None, the reader uses the
default htslib block size.
downsample_fraction: float in the interval [0.0, 1.0] or None. If
specified as a positive float, the reader will only keep each read with
probability downsample_fraction, randomly. If None or zero, all reads
are kept.
random_seed: None or int. The random seed to use with this sam reader, if
needed. If None, a fixed random value will be assigned.
Raises:
ValueError: If downsample_fraction is not None and not in the interval
(0.0, 1.0].
ImportError: If someone tries to load a tfbam file.
"""
if input_path.endswith('.tfbam'):
# Delayed loading of tfbam_lib.
try:
from tfbam_lib import tfbam_reader # pylint: disable=g-import-not-at-top
self._reader = tfbam_reader.make_sam_reader(
input_path,
read_requirements=read_requirements,
unused_block_size=hts_block_size,
downsample_fraction=downsample_fraction,
random_seed=random_seed)
except ImportError:
raise ImportError(
'tfbam_lib module not found, cannot read .tfbam files.')
else:
aux_field_handling = reads_pb2.SamReaderOptions.SKIP_AUX_FIELDS
if parse_aux_fields:
aux_field_handling = reads_pb2.SamReaderOptions.PARSE_ALL_AUX_FIELDS
# We make 0 be a valid value that means "keep all reads" so that proto
# defaults (=0) do not omit all reads.
if downsample_fraction is not None and downsample_fraction != 0:
if not 0.0 < downsample_fraction <= 1.0:
raise ValueError(
'downsample_fraction must be in the interval (0.0, 1.0]',
downsample_fraction)
if random_seed is None:
# Fixed random seed produced with 'od -vAn -N4 -tu4 < /dev/urandom'.
random_seed = 2928130004
self._reader = sam_reader.SamReader.from_file(
input_path.encode('utf8'),
ref_path.encode('utf8') if ref_path is not None else '',
reads_pb2.SamReaderOptions(
read_requirements=read_requirements,
aux_field_handling=aux_field_handling,
hts_block_size=(hts_block_size or 0),
downsample_fraction=downsample_fraction,
random_seed=random_seed))
self.header = self._reader.header
super(NativeSamReader, self).__init__()
def __init__(self,
input_path,
use_index=True,
read_requirements=None,
parse_aux_fields=False,
hts_block_size=None,
downsample_fraction=None,
random_seed=None):
"""Initializes a NativeSamReader.
Args:
input_path: string. A path to a resource containing SAM/BAM records.
Currently supports SAM text format and BAM binary format.
use_index: optional bool, defaulting to True. If True, we will attempt to
load an index file for reads_source to enable the query() API call. If
True an index file must exist. If False, we will not attempt to load an
index for reads_source, disabling the query() call.
read_requirements: optional ReadRequirement proto. If not None, this proto
is used to control which reads are filtered out by the reader before
they are passed to the client.
parse_aux_fields: optional bool. If False, the default, we will not parse
the auxillary fields of the SAM/BAM records (see SAM spec for details).
Parsing the aux fields is often unnecessary for many applications, and
adds a significant parsing cost to access. If you need these aux fields,
set parse_aux_fields to True and these fields will be parsed and
populate the appropriate Read proto fields (e.g., read.info).
hts_block_size: integer or None. If None, will use the default htslib
block size. Otherwise, will configure the underlying block size of the
underlying htslib file object. Larger values (e.g. 1M) may be
beneficial for reading remote files.
downsample_fraction: None or float in the interval [0.0, 1.0]. If not
None or 0.0, the reader will only keep each read with probability
downsample_fraction, randomly.
random_seed: None or int. The random seed to use with this sam reader, if
needed. If None, a fixed random value will be assigned.
Raises:
ValueError: If downsample_fraction is not None and not in the interval
(0.0, 1.0].
ImportError: If someone tries to load a tfbam file.
"""
if input_path.endswith('.tfbam'):
# Delayed loading of tfbam_lib.
try:
from tfbam_lib import tfbam_reader # pylint: disable=g-import-not-at-top
self._reader = tfbam_reader.make_sam_reader(
input_path,
read_requirements=read_requirements,
use_index=use_index,
unused_block_size=hts_block_size,
downsample_fraction=downsample_fraction,
random_seed=random_seed)
except ImportError:
raise ImportError(
'tfbam_lib module not found, cannot read .tfbam files.')
else:
index_mode = index_pb2.INDEX_BASED_ON_FILENAME
if not use_index:
index_mode = index_pb2.DONT_USE_INDEX
aux_field_handling = reads_pb2.SamReaderOptions.SKIP_AUX_FIELDS
if parse_aux_fields:
aux_field_handling = reads_pb2.SamReaderOptions.PARSE_ALL_AUX_FIELDS
if downsample_fraction:
if not 0.0 < downsample_fraction <= 1.0:
raise ValueError(
'downsample_fraction must be in the interval (0.0, 1.0]',
downsample_fraction)
if random_seed is None:
# Fixed random seed produced with 'od -vAn -N4 -tu4 < /dev/urandom'.
random_seed = 2928130004
self._reader = sam_reader.SamReader.from_file(
input_path.encode('utf8'),
reads_pb2.SamReaderOptions(
read_requirements=read_requirements,
index_mode=index_mode,
aux_field_handling=aux_field_handling,
hts_block_size=(hts_block_size or 0),
downsample_fraction=downsample_fraction,
random_seed=random_seed))
self.header = self._reader.header
super(NativeSamReader, self).__init__()
