def create_or_get_subtable(table, subtable_key, size):
subtable = get_in(table, subtable_key)
if subtable is None:
subtable = dict(READGROUP_TEMPLATE)
subtable["Size"] = size
set_in(table, subtable_key, subtable)
return subtable
def update_gtf_table(table, gtf, scaffolds, contig_prefix):
# Workaround for bug in Pysam, which mis-parses individual properties
# (e.g. exon_number) if these are not quoted. This does not apply to
# asDict, which uses a different parsing implementation (v0.7.8).
properties = gtf.asDict()
gene_type = properties.get("gene_biotype")
if gene_type is None:
gene_type = properties.get("gene_type", "unknown_genetype")
keys = (gene_type, properties["gene_id"], properties["transcript_id"],
int(properties["exon_number"]), gtf.feature)
record = {
"contig": contig_prefix + gtf.contig,
"start": gtf.start,
# In pysam, 'end' equals the past-the-end position
"end": gtf.end - 1,
"strand": gtf.strand,
"feature": gtf.feature,
"transcript": properties["transcript_id"]
}
if record["contig"] in scaffolds:
contig = scaffolds[record["contig"]]
record["contig"] = contig["chrom"]
record["start"] += int(contig["chromStart"])
record["end"] += int(contig["chromStart"])
assert not get_in(table, keys), keys
set_in(table, keys, record)
def update_gtf_table(table, gtf, scaffolds, contig_prefix):
# Workaround for bug in Pysam, which mis-parses individual properties
# (e.g. exon_number) if these are not quoted. This does not apply to
# asDict, which uses a different parsing implementation (v0.7.8).
properties = gtf.asDict()
gene_type = properties.get("gene_biotype")
if gene_type is None:
gene_type = properties.get("gene_type", "unknown_genetype")
keys = (gene_type,
properties["gene_id"],
properties["transcript_id"],
int(properties["exon_number"]),
gtf.feature)
record = {"contig": contig_prefix + gtf.contig,
"start": gtf.start,
# In pysam, 'end' equals the past-the-end position
"end": gtf.end - 1,
"strand": gtf.strand,
"feature": gtf.feature,
"transcript": properties["transcript_id"]}
if record["contig"] in scaffolds:
contig = scaffolds[record["contig"]]
record["contig"] = contig["chrom"]
record["start"] += int(contig["chromStart"])
record["end"] += int(contig["chromStart"])
assert not get_in(table, keys), keys
set_in(table, keys, record)
def create_or_get_subtable(table, subtable_key, size):
subtable = get_in(table, subtable_key)
if subtable is None:
subtable = ReadGroup()
subtable.Size = size
set_in(table, subtable_key, subtable)
return subtable
def create_or_get_subtable(table, subtable_key, size):
subtable = get_in(table, subtable_key)
if subtable is None:
subtable = ReadGroup()
subtable.Size = size
set_in(table, subtable_key, subtable)
return subtable
def _collect_clade_from(self, cache, p_node, c_node):
c_clade = get_in(cache, (p_node, c_node), set())
if not c_clade:
if self.is_leaf(c_node):
c_clade.add(c_node)
for n_node in self.connections[c_node]:
if n_node != p_node:
c_clade.update(
self._collect_clade_from(cache, c_node, n_node))
set_in(cache, (p_node, c_node), frozenset(c_clade))
return c_clade
def convert_reads(config, destination, record, sink_cache):
# Source name is used, to re-merge split lanes
name = record.tags.get("PU_src")
destination = os.path.join(destination, name)
make_dirs(os.path.join(config.destination, destination))
def _open_se_sink(reads_type):
key = (name, reads_type)
if not get_in(sink_cache, key):
filename = ReadSink.get_filename(destination, reads_type.lower())
set_in(sink_cache, key, ReadSink.open(config.destination, filename))
return key
for (reads_type, bam_files) in record.bams.iteritems():
# Processed reads are pre-aligned BAMs which have been cleaned up
if reads_type in ("Paired", "Processed"):
# Record "Single" reads; these may result from orphan SE reads
_open_se_sink("Singleton")
key = (name, "Paired")
if not get_in(sink_cache, key):
set_in(sink_cache, key, PEReadSink.open(config.destination,
destination))
else:
key = _open_se_sink(reads_type)
sink = get_in(sink_cache, key)
for filename in bam_files:
print("%sProcessing file %r" % (_INDENTATION * 4, filename))
with pysam.Samfile(filename) as handle:
def _keep_record(record):
return (record.qual >= config.min_quality) and \
(len(record.seq) >= config.min_length)
sink.write_records(record for record in handle
if _keep_record(record))
def convert_reads(config, destination, record, sink_cache):
# Source name is used, to re-merge split lanes
name = record.tags.get("PU_src")
destination = os.path.join(destination, name)
make_dirs(os.path.join(config.destination, destination))
def _open_se_sink(reads_type):
key = (name, reads_type)
if not get_in(sink_cache, key):
filename = ReadSink.get_filename(destination, reads_type.lower())
set_in(sink_cache, key, ReadSink.open(config.destination, filename))
return key
for (reads_type, bam_files) in record.bams.iteritems():
# Processed reads are pre-aligned BAMs which have been cleaned up
if reads_type in ("Paired", "Processed"):
# Record "Single" reads; these may result from orphan SE reads
_open_se_sink("Singleton")
key = (name, "Paired")
if not get_in(sink_cache, key):
set_in(sink_cache, key, PEReadSink.open(config.destination,
destination))
else:
key = _open_se_sink(reads_type)
sink = get_in(sink_cache, key)
for filename in bam_files:
print("%sProcessing file %r" % (_INDENTATION * 4, filename))
with pysam.Samfile(filename) as handle:
def _keep_record(record):
return (record.qual >= config.min_quality) and \
(len(record.seq) >= config.min_length)
sink.write_records(record for record in handle
if _keep_record(record))
def read_table(table, filename):
with open(filename) as table_file:
for record in parse_padded_table(table_file):
key = (record["Name"], record["Sample"],
record["Library"], record["Contig"])
if "*" in key:
continue
subtable = get_in(table, key)
if subtable is None:
subtable = dict(READGROUP_TEMPLATE)
subtable["Size"] = int(record["Size"])
set_in(table, key, subtable)
assert int(subtable["Size"]) == int(record["Size"])
for key in READGROUP_TEMPLATE:
if key != "Size":
subtable[key] += int(record.get(key, 0))
def read_table(table, filename):
with open(filename) as table_file:
for record in parse_padded_table(table_file):
key = (record["Name"], record["Sample"],
record["Library"], record["Contig"])
if "*" in key:
continue
subtable = get_in(table, key)
if subtable is None:
subtable = ReadGroup()
subtable.Size = int(record["Size"])
set_in(table, key, subtable)
assert int(subtable.Size) == int(record["Size"])
for key in ReadGroup.__slots__:
if key != "Size":
subtable[key] += int(record.get(key, 0))
def _read_coverage_tables(cls, key, filenames):
hits = nts = 0
for filename in filenames:
subtable = {}
read_coverage_table(subtable, filename)
contigtables = get_in(subtable, key)
if contigtables is None:
raise NodeError("Error reading table %r; row not found:"
"\n %s ...\n\nIf files have been renamed "
"during the run, then please remove this file "
"in that it may be re-generated.\nHowever, "
"note that read-group tags in the BAM files "
"may not be correct!"
% (filename, " ".join(key)))
for contigtable in contigtables.itervalues():
hits += contigtable["Hits"]
nts += contigtable["M"]
return hits, nts
def read_table(table, filename):
with open(filename) as table_file:
for record in parse_padded_table(table_file):
key = (
record["Name"],
record["Sample"],
record["Library"],
record["Contig"],
)
if "*" in key:
continue
subtable = get_in(table, key)
if subtable is None:
subtable = ReadGroup()
subtable.Size = int(record["Size"])
set_in(table, key, subtable)
assert int(subtable.Size) == int(record["Size"])
for key in ReadGroup.__slots__:
if key != "Size":
subtable[key] += int(record.get(key, 0))
def test_get_in__get_value_two_keywords():
assert utils.get_in({1: {2: 3}}, [1, 2]) == 3
def _open_se_sink(reads_type):
key = (name, reads_type)
if not get_in(sink_cache, key):
filename = ReadSink.get_filename(destination, reads_type.lower())
set_in(sink_cache, key, ReadSink.open(config.destination, filename))
return key
def test_get_in__get_default_three_keywords_fail_at_first():
assert utils.get_in({1: {2: {3: 4}}}, [2, 2, 4]) is None
def test_get_in__get_default_one_keyword():
assert utils.get_in({1: 2}, [2]) is None
def test_get_in__get_value_one_keyword():
assert utils.get_in({1: 2}, [1]) == 2
def _open_se_sink(reads_type):
key = (name, reads_type)
if not get_in(sink_cache, key):
filename = ReadSink.get_filename(destination, reads_type.lower())
set_in(sink_cache, key, ReadSink.open(config.destination, filename))
return key
