def get_or_create_blastdb(blastdb_or_path, dbtype=None, directory=None):
'''it returns a blast database.
If it does not exists it creates and if you give it a directory it will
create it in that directory if it does not exist yet
'''
seq_fpath = _get_abs_blastdb_path(blastdb_or_path, dbtype)
if directory:
dbname = os.path.basename(seq_fpath)
dbpath = os.path.join(directory, dbname)
else:
dbpath = seq_fpath
if not _blastdb_exists(dbpath, dbtype=dbtype):
if not os.path.exists(seq_fpath):
msg = 'An input sequence is required to create a blastdb'
raise RuntimeError(msg)
if seq_fpath != dbpath:
seqio([open(seq_fpath)],
open(dbpath, 'w'),
out_format='fasta',
copy_if_same_format=False)
if dbtype is None:
dbtype = guess_seq_type(open(dbpath))
_makeblastdb_plus(dbpath, dbtype)
return dbpath
def _look_for_blast_matches(self, seq_fpath, oligos, seqs_type):
'It looks for the oligos in the given sequence files'
# we need to keep the blast_fhands, because they're temp files and
# otherwise they might be removed
temp_dir = TemporaryDir()
dbpath = os.path.join(temp_dir.name, os.path.basename(seq_fpath))
seqio([open(seq_fpath)],
open(dbpath, 'w'),
out_format='fasta',
copy_if_same_format=False)
blasts, blast_fhand = _do_blast_2(dbpath,
oligos,
params=self.params,
program=self.program,
dbtype=seqs_type)
if self.filters is not None:
blasts = filter_alignments(blasts, config=self.filters)
# Which are the regions covered in each sequence?
indexed_match_parts = {}
for blast in blasts:
oligo = blast['query']
for match in blast['matches']:
read = match['subject']
if self.elongate_for_global:
elongate_match_parts_till_global(
match['match_parts'],
query_length=oligo['length'],
subject_length=read['length'],
align_completely=QUERY)
# match_parts = [m['match_parts'] for m in blast['matches']]
match_parts = match['match_parts']
try:
indexed_match_parts[read['name']].extend(match_parts)
except KeyError:
indexed_match_parts[read['name']] = match_parts
temp_dir.close()
blast_fhand.close()
return indexed_match_parts
def _look_for_blast_matches(self, seq_fpath, oligos, seqs_type):
'It looks for the oligos in the given sequence files'
# we need to keep the blast_fhands, because they're temp files and
# otherwise they might be removed
temp_dir = TemporaryDir()
dbpath = os.path.join(temp_dir.name, os.path.basename(seq_fpath))
seqio([open(seq_fpath)], open(dbpath, 'w'), out_format='fasta',
copy_if_same_format=False)
blasts, blast_fhand = _do_blast_2(dbpath, oligos, params=self.params,
program=self.program,
dbtype=seqs_type)
if self.filters is not None:
blasts = filter_alignments(blasts, config=self.filters)
# Which are the regions covered in each sequence?
indexed_match_parts = {}
for blast in blasts:
oligo = blast['query']
for match in blast['matches']:
read = match['subject']
if self.elongate_for_global:
elongate_match_parts_till_global(match['match_parts'],
query_length=oligo['length'],
subject_length=read['length'],
align_completely=QUERY)
# match_parts = [m['match_parts'] for m in blast['matches']]
match_parts = match['match_parts']
try:
indexed_match_parts[read['name']].extend(match_parts)
except KeyError:
indexed_match_parts[read['name']] = match_parts
temp_dir.close()
blast_fhand.close()
return indexed_match_parts
def get_or_create_blastdb(blastdb_or_path, dbtype=None, directory=None):
'''it returns a blast database.
If it does not exists it creates and if you give it a directory it will
create it in that directory if it does not exist yet
'''
seq_fpath = _get_abs_blastdb_path(blastdb_or_path, dbtype)
if directory:
dbname = os.path.basename(seq_fpath)
dbpath = os.path.join(directory, dbname)
else:
dbpath = seq_fpath
if not _blastdb_exists(dbpath, dbtype=dbtype):
if not os.path.exists(seq_fpath):
msg = 'An input sequence is required to create a blastdb'
raise RuntimeError(msg)
if seq_fpath != dbpath:
seqio([open(seq_fpath)], open(dbpath, 'w'), out_format='fasta',
copy_if_same_format=False)
if dbtype is None:
dbtype = guess_seq_type(open(dbpath))
_makeblastdb_plus(dbpath, dbtype)
return dbpath
def test_seqio(self):
'It tets the seqio function'
# fastq to fasta
out_fhand = NamedTemporaryFile()
seqio([self._make_fhand(FASTQ)], out_fhand, 'fasta')
assert ">seq1\natcgt" in open(out_fhand.name).read()
# fastq to fastq-illumina
out_fhand = NamedTemporaryFile()
seqio([self._make_fhand(FASTQ)], out_fhand, 'fastq-illumina')
assert "@seq1\natcgt\n+\n^^^^" in open(out_fhand.name).read()
out_fhand = NamedTemporaryFile()
seqio([self._make_fhand(FASTQ),
self._make_fhand(FASTQ)], out_fhand, 'fastq-illumina')
assert "@seq3\natcgt\n+\n^^^^^\n@seq1" in open(out_fhand.name).read()
# fasta to fastq
out_fhand = NamedTemporaryFile()
try:
seqio([self._make_fhand(FASTA)], out_fhand, 'fastq')
self.fail("error previously expected")
except IncompatibleFormatError as error:
assert 'No qualities available' in str(error)
# bad_format fastq
bad_fastq_fhand = self._make_fhand(FASTQ + 'aklsjhdas')
try:
seqio([bad_fastq_fhand], out_fhand, 'fasta')
self.fail("error previously expected")
except MalformedFile as error:
assert 'Lengths of sequence and quality' in str(error)
# genbank to fasta
out_fhand = NamedTemporaryFile()
genbank_fhand = open(os.path.join(TEST_DATA_DIR, 'sequence.gb'))
seqio([genbank_fhand], out_fhand, 'fasta')
result = open(out_fhand.name).read()
assert '>NM_019354.2' in result
def test_seqio(self):
'It tets the seqio function'
# fastq to fasta
out_fhand = NamedTemporaryFile()
seqio([self._make_fhand(FASTQ)], out_fhand, 'fasta')
assert ">seq1\natcgt" in open(out_fhand.name).read()
# fastq to fastq-illumina
out_fhand = NamedTemporaryFile()
seqio([self._make_fhand(FASTQ)], out_fhand, 'fastq-illumina')
assert "@seq1\natcgt\n+\n^^^^" in open(out_fhand.name).read()
out_fhand = NamedTemporaryFile()
seqio([self._make_fhand(FASTQ), self._make_fhand(FASTQ)],
out_fhand, 'fastq-illumina')
assert "@seq3\natcgt\n+\n^^^^^\n@seq1" in open(out_fhand.name).read()
# fasta to fastq
out_fhand = NamedTemporaryFile()
try:
seqio([self._make_fhand(FASTA)], out_fhand, 'fastq')
self.fail("error previously expected")
except IncompatibleFormatError as error:
assert 'No qualities available' in str(error)
# bad_format fastq
bad_fastq_fhand = self._make_fhand(FASTQ + 'aklsjhdas')
try:
seqio([bad_fastq_fhand], out_fhand, 'fasta')
self.fail("error previously expected")
except MalformedFile as error:
assert 'Lengths of sequence and quality' in str(error)
# genbank to fasta
out_fhand = NamedTemporaryFile()
genbank_fhand = open(os.path.join(TEST_DATA_DIR, 'sequence.gb'))
seqio([genbank_fhand], out_fhand, 'fasta')
result = open(out_fhand.name).read()
assert '>NM_019354.2' in result
