def setup(self):
thisdir = os.path.dirname(__file__)
self._fileName = os.path.join(thisdir, 'fastqRecovery')
self._testfq = os.path.join(thisdir, 'test.fastq')
screed.read_fastq_sequences(self._testfq)
screed.ToFastq(self._testfq, self._fileName)
screed.read_fastq_sequences(self._fileName)
self.db = screed.ScreedDB(self._fileName)
def setup(self):
thisdir = os.path.dirname(__file__)
self._fileName = os.path.join(thisdir, 'fastqRecovery')
self._testfq = os.path.join(thisdir, 'test.fastq')
screed.read_fastq_sequences(self._testfq)
screed.ToFastq(self._testfq, self._fileName)
screed.read_fastq_sequences(self._fileName)
self.db = screed.ScreedDB(self._fileName)
def setup(self):
self._fileName = utils.get_temp_filename('fastqRecovery')
self._testfq = utils.get_temp_filename('test.fastq')
shutil.copy(utils.get_test_data('test.fastq'), self._testfq)
screed.read_fastq_sequences(self._testfq)
screed.ToFastq(self._testfq, self._fileName)
screed.read_fastq_sequences(self._fileName)
self.db = screed.ScreedDB(self._fileName)
def setup(self):
self._fileName = utils.get_temp_filename('fastqRecovery')
self._testfq = utils.get_temp_filename('test.fastq')
shutil.copy(utils.get_test_data('test.fastq'), self._testfq)
screed.read_fastq_sequences(self._testfq)
screed.ToFastq(self._testfq, self._fileName)
screed.read_fastq_sequences(self._fileName)
self.db = screed.ScreedDB(self._fileName)
def setup(self):
thisdir = os.path.dirname(__file__)
self._fqName = os.path.join(thisdir, 'fa_to_fq')
self._faName = os.path.join(thisdir, 'fq_to_fa')
self._testfa = os.path.join(thisdir, 'test.fa')
screed.read_fasta_sequences(self._testfa)
screed.ToFastq(self._testfa, self._fqName) # Fasta db -> fasta text
screed.read_fastq_sequences(self._fqName) # Fastq file -> fastq db
screed.ToFasta(self._fqName, self._faName) # Fastq db -> fasta text
screed.read_fasta_sequences(self._faName) # Fasta file -> fasta db
self.db = screed.ScreedDB(self._faName)
def setup(self):
self._fqName = utils.get_temp_filename('fa_to_fq')
self._faName = utils.get_temp_filename('fq_to_fa')
self._testfa = utils.get_temp_filename('test.fa')
shutil.copy(utils.get_test_data('test.fa'), self._testfa)
screed.read_fasta_sequences(self._testfa)
screed.ToFastq(self._testfa, self._fqName) # Fasta db -> fasta text
screed.read_fastq_sequences(self._fqName) # Fastq file -> fastq db
screed.ToFasta(self._fqName, self._faName) # Fastq db -> fasta text
screed.read_fasta_sequences(self._faName) # Fasta file -> fasta db
self.db = screed.ScreedDB(self._faName)
def setup(self):
self._fqName = utils.get_temp_filename('fa_to_fq')
self._faName = utils.get_temp_filename('fq_to_fa')
self._testfa = utils.get_temp_filename('test.fa')
shutil.copy(utils.get_test_data('test.fa'), self._testfa)
screed.read_fasta_sequences(self._testfa)
screed.ToFastq(self._testfa, self._fqName) # Fasta db -> fasta text
screed.read_fastq_sequences(self._fqName) # Fastq file -> fastq db
screed.ToFasta(self._fqName, self._faName) # Fastq db -> fasta text
screed.read_fasta_sequences(self._faName) # Fasta file -> fasta db
self.db = screed.ScreedDB(self._faName)
def time_screed(f, random_seqs, name):
show_name(name)
rm("%s_%s" % (f, screed.DBConstants.fileExtension))
t = time.time()
screed.read_fastq_sequences(f)
print "create: %.3f" % (time.time() - t)
faqdb = screed.ScreedDB(f)
t = time.time()
for r in random_seqs:
faqdb[r[1:]].sequence
print "search: %.3f" % (time.time() - t)
del faqdb
def time_screed(f, random_seqs, name):
show_name(name)
rm("%s_%s" % (f, screed.DBConstants.fileExtension))
t = time.time()
screed.read_fastq_sequences(f)
print "create: %.3f" % (time.time() - t)
faqdb = screed.ScreedDB(f)
t = time.time()
for r in random_seqs:
faqdb[r[1:]].sequence
print "search: %.3f" % (time.time() - t)
del faqdb
def openDB(fileName):
"""Opening screed DB; making if not already existing
Args:
fileName -- Name of sequence file or screedDB file
"""
logging.info('{}: Making/opening screed database for: "{}"'.format(my_time(), fileName))
# making db if needed
if not fileName.endswith('_screed'):
try:
screed.read_fastq_sequences(fileName)
fileName = fileName + '_screed'
except KeyError:
try:
screed.read_fasta_sequences(fileName)
fileName = fileName + '_screed'
except IOError:
msg = 'Cannot open {}'.format(fileName)
raise IOError(msg)
# init screed db
return screed.ScreedDB(fileName)
def setup(self):
self._testfq = os.path.join(os.path.dirname(__file__), 'test.fastq')
screed.read_fastq_sequences(self._testfq)
self.db = screed.ScreedDB(self._testfq)
def setup(self):
self._testfq = os.path.join(os.path.dirname(__file__), 'test.fastq')
screed.read_fastq_sequences(self._testfq)
self.db = screed.ScreedDB(self._testfq)
python both.py R1.fastq R2.fastq
out put:
R1.fastq.both
R2.fastq.both
This script uses the screed module: https://github.com/ctb/screed
'''
import screed
import sys
R1_IN = sys.argv[1]
R2_IN = sys.argv[2]
screed.read_fastq_sequences(R1_IN)
screed.read_fastq_sequences(R2_IN)
DB_R1 = screed.ScreedDB(R1_IN+'_screed')
DB_R2 = screed.ScreedDB(R2_IN+'_screed')
with open(R1_IN+'.both','w') as R1_OUT:
with open(R2_IN+'.both','w') as R2_OUT:
for record, thing in DB_R1.iteritems():
try:
match = DB_R2[thing['name'].replace(" 1:"," 2:")]
except KeyError:
continue
R1_OUT.write('@%s %s\n%s\n+\n%s\n' % (thing['name'],thing['annotations'],thing['sequence'],thing['accuracy']))
R2_OUT.write('@%s %s\n%s\n+\n%s\n' % (match['name'],match['annotations'],match['sequence'],match['accuracy']))
screeddb = Filein + '_screed'
if os.path.exists(screeddb):
db = screed.ScreedDB(screeddb)
if Percent != 100 and os.path.exists(os.path.join(Mypath,'temp')):
print("Screed database and shuffled files already exists proceeding to resampling.")
resamp(Percent, os.path.join(Mypath,'temp'), Mypath, Fileout, db, sys.argv[3], NumResamp)
os.remove(os.path.join(Mypath,'temp'))
print('Shuffleing done with resampleing. No log file written.')
sys.exit()
else:
print("Screed database already exists proceeding to shuffleing.")
else:
print("Creating screed database.")
db = screed.read_fastq_sequences(Filein)
print(dict(db))
ReadNum = len(db)
N='name'
S='sequence'
A='accuracy'
n, j, k, l = 0, 0, 0, 0
with open(Fileout, 'w') as outfile:
with open(Singleton1, 'w') as R1:
with open(Singleton2, 'w') as R2:
for record, thing in db.iteritems():
#!/usr/bin/env python
# Copyright (c) 2008-2010, Michigan State University
from __future__ import absolute_import
import sys
from screed import read_fastq_sequences
from screed import DBConstants
# A shell interface to the screed FQDBM database writing function
if __name__ == "__main__":
# Make sure the user entered the command line arguments correctly
if len(sys.argv) != 2:
sys.stderr.write("ERROR: USAGE IS: %s <dbfilename>\n" % sys.argv[0])
exit(1)
filename = sys.argv[1]
read_fastq_sequences(filename)
print("Database saved in %s%s" % (sys.argv[1], DBConstants.fileExtension))
exit(0)