def setup_maf_header(self):
"""
Sets up the maf header.
"""
# Reader header
_hdr = MafHeader.from_reader(reader=self.maf_reader)
if not self.options["reference_fasta_index"]:
self.maf_header = MafHeader.from_defaults(
version=self.options["version"],
annotation=self.options["annotation"],
sort_order=BarcodesAndCoordinate(),
contigs=_hdr.contigs(),
)
else:
self.maf_header = MafHeader.from_defaults(
version=self.options["version"],
annotation=self.options["annotation"],
sort_order=BarcodesAndCoordinate(),
fasta_index=self.options["reference_fasta_index"],
)
self.maf_header.validation_stringency = ValidationStringency.Strict
header_date = BaseRunner.get_header_date()
self.maf_header[header_date.key] = header_date
try:
nkey = _hdr["normal.aliquot"]
self.maf_header["normal.aliquot"] = nkey
except KeyError as e:
if not self.options["tumor_only"]:
raise e
tkey = _hdr["tumor.aliquot"]
self.maf_header["tumor.aliquot"] = tkey
def setup_maf_header(self):
"""
Sets up the maf header.
"""
self.maf_header = MafHeader.from_defaults(
version=self.options["version"],
annotation=self.options["annotation"],
sort_order=BarcodesAndCoordinate(),
fasta_index=self.options["reference_fasta_index"],
)
header_date = BaseRunner.get_header_date()
self.maf_header[header_date.key] = header_date
if not self.options["tumor_only"]:
normal_aliquot = MafHeaderRecord(
key="normal.aliquot",
value=self.options["normal_aliquot_uuid"]
if not self.options["tumor_only"]
else "",
)
self.maf_header[normal_aliquot.key] = normal_aliquot
tumor_aliquot = MafHeaderRecord(
key="tumor.aliquot", value=self.options["tumor_aliquot_uuid"]
)
self.maf_header[tumor_aliquot.key] = tumor_aliquot
def __init__(self,
iters,
fasta_index=None,
contigs=None,
by_barcodes=True,
peekable_iterator_class=PeekableIterator):
"""
:param iters: the list of iterators.
:param fasta_index: the path to the FASTA index for defining
ordering across chromosomes.
:param contigs: the list of contigs to use for sorting instead of
parsing from FASTA index.
:param by_barcodes: True to require the same tumor and matched
normal barcodes for returned locatables, False otherwise
:param peekable_iterator_class: PeekableIterator class to use when
traversing individual MAFs. This allows developers to add in custom
filters and custom handling of MAFs.
"""
self._by_barcodes = by_barcodes
if self._by_barcodes:
self._sort_order = BarcodesAndCoordinate(fasta_index=fasta_index, contigs=contigs)
self._overlap_f = self.__overlaps_with_barcode
else:
self._sort_order = Coordinate(fasta_index=fasta_index)
self._overlap_f = self.__overlaps
# Trust, but verify
_iters = [_SortOrderEnforcingIterator(_iter, self._sort_order)
for _iter in iters]
self._iters = [peekable_iterator_class(_iter) for _iter in _iters]
self._sort_key = self._sort_order.sort_key()
def __init__(
self,
iters: List[MafReader],
fasta_index: Optional[str] = None,
contigs: List[str] = None,
by_barcodes: bool = True,
peekable_iterator_class: Type[PeekableIterator] = PeekableIterator,
):
"""
:param iters: the list of iterators.
:param fasta_index: the path to the FASTA index for defining
ordering across chromosomes.
:param contigs: the list of contigs to use for sorting instead of
parsing from FASTA index.
:param by_barcodes: True to require the same tumor and matched
normal barcodes for returned locatables, False otherwise
:param peekable_iterator_class: PeekableIterator class to use when
traversing individual MAFs. This allows developers to add in custom
filters and custom handling of MAFs.
"""
self._overlap_f: Union[Callable[
[_BarcodesAndCoordinateKey, _BarcodesAndCoordinateKey], bool],
Callable[[Locatable, Locatable], bool], ]
if not by_barcodes:
_sort_order = Coordinate(fasta_index=fasta_index)
self._overlap_f = self.__overlaps
else:
_sort_order = BarcodesAndCoordinate(fasta_index=fasta_index,
contigs=contigs)
self._overlap_f = self.__overlaps_with_barcode
self._sort_order: Coordinate = _sort_order
self._by_barcodes: bool = by_barcodes
# Trust, but verify
_iters = [
_SortOrderEnforcingIterator(_iter, self._sort_order)
for _iter in iters
]
self._iters: List[PeekableIterator] = [
peekable_iterator_class(_iter) for _iter in _iters
]
self._sort_key: TSortKey = self._sort_order.sort_key()
def test_less_than_max_in_memory(self):
max_objects_in_ram = 100
num_records = max_objects_in_ram - 1
sorter = Sorter(max_objects_in_ram,
self.codec(),
BarcodesAndCoordinate().sort_key(),
always_spill=False)
# add them in reverse order
for i in range(num_records):
record = DummyRecord("A", "B", "C", 1, num_records - i - 1)
sorter += record
records = [r for r in sorter]
sorter.close()
self.assertEqual(len(records), num_records)
for i in range(num_records):
record = records[i]
self.assertEqual(record.value("End_Position"), i)
def setup_maf_header(self):
"""
Sets up the maf header.
"""
# Reader header
_hdr = MafHeader.from_reader(reader=self.maf_readers[0])
self.maf_header = MafHeader.from_defaults(
version=self.options['version'],
annotation=self.options['annotation'],
sort_order=BarcodesAndCoordinate(),
contigs=_hdr.contigs())
self.maf_header.validation_stringency = ValidationStringency.Strict
header_date = BaseRunner.get_header_date()
self.maf_header[header_date.key] = header_date
nkey = _hdr["normal.aliquot"]
self.maf_header["normal.aliquot"] = nkey
tkey = _hdr["tumor.aliquot"]
self.maf_header["tumor.aliquot"] = tkey
def test_empty(self):
sorter = Sorter(100, self.codec(), BarcodesAndCoordinate().sort_key())
records = [r for r in sorter]
sorter.close()
self.assertEqual(len(records), 0)