def _run(self, _config, temp):
def keyfunc(bed):
return (bed.contig, bed.name, bed.start)
fastafile = pysam.Fastafile(self._reference)
seqs = collections.defaultdict(list)
with open(self._intervals) as bedfile:
intervals = text.parse_lines_by_contig(bedfile, pysam.asBed()).items()
for (contig, beds) in sorted(intervals):
beds.sort(key = keyfunc)
for (gene, gene_beds) in itertools.groupby(beds, lambda x: x.name):
gene_beds = tuple(gene_beds)
for bed in gene_beds:
seqs[(contig, gene)].append(fastafile.fetch(contig, bed.start, bed.end))
seq = "".join(seqs[(contig, gene)])
if any((bed.strand == "-") for bed in gene_beds):
assert all((bed.strand == "-") for bed in gene_beds)
seq = sequences.reverse_complement(seq)
seqs[(contig, gene)] = seq
temp_file = os.path.join(temp, "sequences.fasta")
with open(temp_file, "w") as out_file:
for ((_, gene), sequence) in sorted(seqs.items()):
fasta.print_fasta(gene, sequence, out_file)
move_file(temp_file, self._outfile)
def _run(self, _config, temp):
def keyfunc(bed):
return (bed.contig, bed.name, bed.start)
fastafile = pysam.Fastafile(self._reference)
seqs = collections.defaultdict(list)
with open(self._intervals) as bedfile:
intervals = text.parse_lines_by_contig(bedfile,
pysam.asBed()).items()
for (contig, beds) in sorted(intervals):
beds.sort(key=keyfunc)
for (gene,
gene_beds) in itertools.groupby(beds, lambda x: x.name):
gene_beds = tuple(gene_beds)
for bed in gene_beds:
seqs[(contig, gene)].append(
fastafile.fetch(contig, bed.start, bed.end))
seq = "".join(seqs[(contig, gene)])
if any((bed.strand == "-") for bed in gene_beds):
assert all((bed.strand == "-") for bed in gene_beds)
seq = sequences.reverse_complement(seq)
seqs[(contig, gene)] = seq
temp_file = os.path.join(temp, "sequences.fasta")
with open(temp_file, "w") as out_file:
for ((_, gene), sequence) in sorted(seqs.items()):
fasta.print_fasta(gene, sequence, out_file)
move_file(temp_file, self._outfile)
def test_print_fasta__multiple_lines():
expected = ">foobar\n%s\n%s\n" \
% (_SEQ_FRAG * 10, _SEQ_FRAG * 5)
stringf = StringIO.StringIO()
print_fasta("foobar", _SEQ_FRAG * 15, stringf)
assert_equal(stringf.getvalue(), expected)
def test_print_fasta__complete_line_test():
expected = ">barfoo\n%s\n" % (_SEQ_FRAG * 10, )
stringf = StringIO.StringIO()
print_fasta("barfoo", _SEQ_FRAG * 10, stringf)
assert_equal(stringf.getvalue(), expected)
def test_print_fasta__partial_line():
expected = ">foobar\n%s\n" % (_SEQ_FRAG, )
stringf = StringIO.StringIO()
print_fasta("foobar", _SEQ_FRAG, stringf)
assert_equal(stringf.getvalue(), expected)
def test_print_fasta__multiple_lines():
expected = ">foobar\n%s\n%s\n" \
% (_SEQ_FRAG * 10, _SEQ_FRAG * 5)
stringf = StringIO.StringIO()
print_fasta("foobar", _SEQ_FRAG * 15, stringf)
assert_equal(stringf.getvalue(), expected)
def test_print_fasta__complete_line_test():
expected = ">barfoo\n%s\n" % (_SEQ_FRAG * 10, )
stringf = StringIO.StringIO()
print_fasta("barfoo", _SEQ_FRAG * 10, stringf)
assert_equal(stringf.getvalue(), expected)
def test_print_fasta__partial_line():
expected = ">foobar\n%s\n" % (_SEQ_FRAG, )
stringf = StringIO.StringIO()
print_fasta("foobar", _SEQ_FRAG, stringf)
assert_equal(stringf.getvalue(), expected)
def print_msa(msa, file = sys.stdout):
validate_msa(msa)
for group in sorted(msa):
print_fasta(group, msa[group], file)
def print_msa(msa, file=sys.stdout):
validate_msa(msa)
for group in sorted(msa):
print_fasta(group, msa[group], file)