def main(argv=None):
parser = make_parser()
args = parser.parse_args(argv)
genome = SequenceFileDB(args.genome)
pwm = [IUPAC_SCORES[l] for l in args.consensus]
pwm.extend([REQUIRED_SCORES[l] for l in args.required_3p_seq])
pwm = motility.PWM(pwm)
# find all matches
with open(args.outfile, 'w') as outfile:
for chrom in genome.keys():
chromseq = str(genome[chrom])
print "searching ", chrom, "of length", len(chromseq)
if len(chromseq) < len(pwm):
print 'chromosome/fragment', chrom, 'is too short'
continue
matches = pwm.find(chromseq, -args.mismatches)
for start, stop, strand, seq in matches:
score = pwm.calc_score(seq)
outfile.write('\t'.join(
[chrom, str(start), str(stop), seq, str(score),
'+' if strand == 1 else '-']) + '\n')
class SequenceFileDB_Test(unittest.TestCase):
"""
Test for all of the basic dictionary functions on 'SequenceFileDB',
among other things.
"""
def setUp(self):
"Test setup"
dnaseq = testutil.datafile('dnaseq.fasta')
self.db = SequenceFileDB(dnaseq) # contains 'seq1', 'seq2'
self.db._weakValueDict.clear() # clear the cache
def tearDown(self):
self.db.close() # must close SequenceFileDB!
def test_len(self):
assert len(self.db) == 2
def test_seqInfoDict_len(self):
assert len(self.db.seqInfoDict) == 2
def test_no_file_given(self):
"Make sure that a TypeError is raised when no file is available"
try:
db = SequenceFileDB()
assert 0, "should not reach this point"
except TypeError:
pass
def test_seq_descriptor(self):
"Check the '.seq' attribute (tied to a descriptor)"
s = self.db['seq1']
assert str(s) == str(s.seq)
def test_cache(self):
"SequenceDB cache test"
assert len(self.db._weakValueDict) == 0
seq1 = self.db['seq1']
# cache populated?
assert len(self.db._weakValueDict) == 1
assert 'seq1' in self.db._weakValueDict
# cache functions?
seq1_try2 = self.db['seq1']
assert seq1 is seq1_try2
def test_clear_cache(self):
"SequenceDB clear_cache test"
assert len(self.db._weakValueDict) == 0
seq1 = self.db['seq1']
# cache populated?
assert len(self.db._weakValueDict) == 1
assert 'seq1' in self.db._weakValueDict
# clear_cache functions?
self.db.clear_cache()
seq1_try3 = self.db['seq1']
assert seq1 is not seq1_try3
def test_keys(self):
"SequenceFileDB keys"
k = self.db.keys()
k.sort()
assert k == ['seq1', 'seq2']
def test_contains(self):
"SequenceFileDB contains"
assert 'seq1' in self.db, self.db.keys()
assert 'seq2' in self.db
assert 'foo' not in self.db
def test_invert_class(self):
"SequenceFileDB __invert__"
seq = self.db['seq1']
inversedb = ~self.db
assert inversedb[seq] == 'seq1'
assert seq in inversedb
assert 'foo' not in inversedb
def test_keys_info(self):
"SequenceFileDB keys info"
k = self.db.seqInfoDict.keys()
k.sort()
assert k == ['seq1', 'seq2']
def test_contains_info(self):
"SequenceFileDB contains info"
assert 'seq1' in self.db.seqInfoDict
assert 'seq2' in self.db.seqInfoDict
assert 'foo' not in self.db.seqInfoDict
def test_has_key(self):
"SequenceFileDB has key"
assert 'seq1' in self.db
assert 'seq2' in self.db
assert 'foo' not in self.db
def test_get(self):
"SequenceFileDB get"
assert self.db.get('foo') is None
assert self.db.get('seq1') is not None
assert str(self.db.get('seq1')).startswith('atggtgtca')
assert self.db.get('seq2') is not None
assert str(self.db.get('seq2')).startswith('GTGTTGAA')
def test_items(self):
"SequenceFileDB items"
i = [k for (k, v) in self.db.items()]
i.sort()
assert i == ['seq1', 'seq2']
def test_iterkeys(self):
"SequenceFileDB iterkeys"
kk = self.db.keys()
kk.sort()
ik = list(self.db.iterkeys())
ik.sort()
assert kk == ik
def test_itervalues(self):
"SequenceFileDB itervalues"
kv = self.db.values()
kv.sort()
iv = list(self.db.itervalues())
iv.sort()
assert kv == iv
def test_iteritems(self):
"SequenceFileDB iteritems"
ki = self.db.items()
ki.sort()
ii = list(self.db.iteritems())
ii.sort()
assert ki == ii
def test_readonly(self):
"SequenceFileDB readonly"
try:
self.db.copy() # what should 'copy' do on SequenceFileDB?
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.clear()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.setdefault('foo')
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.pop()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.popitem()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.update({})
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
# test some things other than dict behavior
def test_keyerror(self):
"""SequenceFileDB keyerror.
Make sure that the SequenceFileDB KeyError is informative."""
try:
self.db['foo']
except KeyError, e:
assert "no key 'foo' in database <SequenceFileDB" in str(e), str(e)
class Hgvs2:
def __init__(self, refseq_fn=fileconfig.FILECONFIG['REFGENE']):
""" register hg19 reference genome sequence and NCBI RefSeq transcript coordinates"""
self.genome_fn = fileconfig.FILECONFIG['REFGENOME_UCSC']
self.genome = None
if not os.path.exists(self.genome_fn):
raise IOError(
'Reference Genome Sequence (UCSC format) for %s is not found' %
self.genome_fn)
self.refseq_fn = refseq_fn
self.refseq = None
if not os.path.exists(self.refseq_fn):
raise IOError('NCBI RefSeq transcript for %s is not found' %
self.refseq_fn)
def get_transcript(self, name):
if self.refseq:
return self.refseq.get(name)
else:
raise RuntimeError(
'first, load a transcript coordinate file![%s]' %
self.refseq_fn)
def load_resource(self):
#load genome sequence
print 'loading the genome sequence [%s] for HGVS...' % self.genome_fn
self.genome = SequenceFileDB(self.genome_fn)
print 'done.'
#load refseq into dic
print 'loading the refseq transcript [%s] for HGVS...' % self.refseq_fn
fp = open(self.refseq_fn, 'r')
self.refseq = pyhgvs.utils.read_transcripts(fp)
fp.close()
print 'done.'
def close_resource(self):
self.genome.close()
self.refseq = None
def to_cDNA(self, chrom, offset, ref, alt, refseq_acc):
""" convert to HGVS nomenclature """
transcript = self.get_transcript(refseq_acc)
if not chrom.startswith('chr'):
chrom = 'chr%s' % chrom
if chrom not in CHROMOSOMES:
return None
if not chrom in self.genome.keys():
return None
cdna = pyhgvs.format_hgvs_name(chrom, offset, ref, alt, self.genome,
transcript)
if cdna:
itms = cdna.split(':')
if len(itms) > 1:
cdna = itms[1]
else:
cdna = cdna
return cdna
else:
return None
def gdna_to_vcf(self, gdna):
return pyhgvs.gdna_to_vcf(gdna, self.genome)
def to_chrom_coordinate(self, cDNA):
try:
chrom, offset, ref, alt = pyhgvs.parse_hgvs_name(cDNA, self.genome, \
get_transcript = self.get_transcript)
return chrom, offset, ref, alt
except:
print "[%s] cannot be coverted to chromosome coordinate" % cDNA
return None, None, None, None
class SequenceFileDB_Test(object):
"""
Test for all of the basic dictionary functions on 'SequenceFileDB'.
"""
def setup(self):
self.db = SequenceFileDB('dnaseq') # contains 'seq1', 'seq2'
def keys_test(self):
k = self.db.keys()
k.sort()
assert k == ['seq1', 'seq2']
def contains_test(self):
assert 'seq1' in self.db, self.db.keys()
assert 'seq2' in self.db
assert 'foo' not in self.db
def keys_info_test(self):
k = self.db.seqInfoDict.keys()
k.sort()
assert k == ['seq1', 'seq2']
def contains_info_test(self):
assert 'seq1' in self.db.seqInfoDict
assert 'seq2' in self.db.seqInfoDict
assert 'foo' not in self.db.seqInfoDict
def has_key_test(self):
assert self.db.has_key('seq1')
assert self.db.has_key('seq2')
assert not self.db.has_key('foo')
def get_test(self):
assert self.db.get('foo') is None
assert self.db.get('seq1') is not None
assert str(self.db.get('seq1')).startswith('atggtgtca')
assert self.db.get('seq2') is not None
assert str(self.db.get('seq2')).startswith('GTGTTGAA')
def items_test(self):
i = [ k for (k,v) in self.db.items() ]
i.sort()
assert i == ['seq1', 'seq2']
def iterkeys_test(self):
kk = self.db.keys()
kk.sort()
ik = list(self.db.iterkeys())
ik.sort()
assert kk == ik
def itervalues_test(self):
kv = self.db.values()
kv.sort()
iv = list(self.db.itervalues())
iv.sort()
assert kv == iv
def iteritems_test(self):
ki = self.db.items()
ki.sort()
ii = list(self.db.iteritems())
ii.sort()
assert ki == ii
def readonly_test(self):
try:
self.db.copy() # what should 'copy' do on SequenceFileDB?
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.clear()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.setdefault('foo')
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.pop()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.popitem()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.update({})
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
# test some things other than dict behavior
def keyerror_test(self):
"Make sure that the SequenceFileDB KeyError is informative."
try:
self.db['foo']
except KeyError, e:
assert "no key 'foo' in database <SequenceFileDB" in str(e), str(e)
class SequenceFileDB_Test(unittest.TestCase):
"""
Test for all of the basic dictionary functions on 'SequenceFileDB',
among other things.
"""
def setUp(self):
"Test setup"
dnaseq = testutil.datafile('dnaseq.fasta')
self.db = SequenceFileDB(dnaseq) # contains 'seq1', 'seq2'
self.db._weakValueDict.clear() # clear the cache
def tearDown(self):
self.db.close() # must close SequenceFileDB!
def test_len(self):
assert len(self.db) == 2
def test_seqInfoDict_len(self):
assert len(self.db.seqInfoDict) == 2
def test_no_file_given(self):
"Make sure that a TypeError is raised when no file is available"
try:
db = SequenceFileDB()
assert 0, "should not reach this point"
except TypeError:
pass
def test_seq_descriptor(self):
"Check the '.seq' attribute (tied to a descriptor)"
s = self.db['seq1']
assert str(s) == str(s.seq)
def test_cache(self):
"SequenceDB cache test"
assert len(self.db._weakValueDict) == 0
seq1 = self.db['seq1']
# cache populated?
assert len(self.db._weakValueDict) == 1
assert 'seq1' in self.db._weakValueDict
# cache functions?
seq1_try2 = self.db['seq1']
assert seq1 is seq1_try2
def test_clear_cache(self):
"SequenceDB clear_cache test"
assert len(self.db._weakValueDict) == 0
seq1 = self.db['seq1']
# cache populated?
assert len(self.db._weakValueDict) == 1
assert 'seq1' in self.db._weakValueDict
# clear_cache functions?
self.db.clear_cache()
seq1_try3 = self.db['seq1']
assert seq1 is not seq1_try3
def test_keys(self):
"SequenceFileDB keys"
k = self.db.keys()
k.sort()
assert k == ['seq1', 'seq2']
def test_contains(self):
"SequenceFileDB contains"
assert 'seq1' in self.db, self.db.keys()
assert 'seq2' in self.db
assert 'foo' not in self.db
def test_invert_class(self):
"SequenceFileDB __invert__"
seq = self.db['seq1']
inversedb = ~self.db
assert inversedb[seq] == 'seq1'
assert seq in inversedb
assert 'foo' not in inversedb
def test_keys_info(self):
"SequenceFileDB keys info"
k = self.db.seqInfoDict.keys()
k.sort()
assert k == ['seq1', 'seq2']
def test_contains_info(self):
"SequenceFileDB contains info"
assert 'seq1' in self.db.seqInfoDict
assert 'seq2' in self.db.seqInfoDict
assert 'foo' not in self.db.seqInfoDict
def test_has_key(self):
"SequenceFileDB has key"
assert 'seq1' in self.db
assert 'seq2' in self.db
assert 'foo' not in self.db
def test_get(self):
"SequenceFileDB get"
assert self.db.get('foo') is None
assert self.db.get('seq1') is not None
assert str(self.db.get('seq1')).startswith('atggtgtca')
assert self.db.get('seq2') is not None
assert str(self.db.get('seq2')).startswith('GTGTTGAA')
def test_items(self):
"SequenceFileDB items"
i = [k for (k, v) in self.db.items()]
i.sort()
assert i == ['seq1', 'seq2']
def test_iterkeys(self):
"SequenceFileDB iterkeys"
kk = self.db.keys()
kk.sort()
ik = list(self.db.iterkeys())
ik.sort()
assert kk == ik
def test_itervalues(self):
"SequenceFileDB itervalues"
kv = self.db.values()
kv.sort()
iv = list(self.db.itervalues())
iv.sort()
assert kv == iv
def test_iteritems(self):
"SequenceFileDB iteritems"
ki = self.db.items()
ki.sort()
ii = list(self.db.iteritems())
ii.sort()
assert ki == ii
def test_readonly(self):
"SequenceFileDB readonly"
try:
self.db.copy() # what should 'copy' do on SequenceFileDB?
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.clear()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.setdefault('foo')
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.pop()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.popitem()
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
try:
self.db.update({})
assert 0, 'this method should raise NotImplementedError'
except NotImplementedError:
pass
# test some things other than dict behavior
def test_keyerror(self):
"""SequenceFileDB keyerror.
Make sure that the SequenceFileDB KeyError is informative."""
try:
self.db['foo']
except KeyError, e:
assert "no key 'foo' in database <SequenceFileDB" in str(e), str(e)
