import os, sys
from mungo.fasta import FastaFile
from optparse import OptionParser
usage = "%prog <Input file1> <Output file>"
parser = OptionParser(usage=usage, version="%prog - Version 1")
parser.add_option('-i', '--stdin', dest='stdin', action='store_true', help='Input from stdin')
parser.add_option('-o', '--stdout', dest='stdout', action='store_true', help='Output to stdout')
options, args = parser.parse_args()
if len(args)==0:
parser.print_help()
sys.exit()
if options.stdin:
f = FastaFile(sys.stdin)
else:
f = FastaFile(args[0])
if options.stdout:
w = FastaFile(sys.stdout, 'w')
else:
w = FastaFile(args[1], 'w')
accList= []
for h,s in f:
acc = h.split()[0]
if not acc in accList:
w.write(h,s)
accList.append(acc)
w.close()
fasta_block_split.py <iFilenames> <oFilename>
Author: Tony Papenfuss
Date: Fri Mar 30 12:58:58 EST 2007
"""
import os, sys
from mungo.fasta import FastaFile
from optparse import OptionParser
usage = "%prog [-b <block_size>] <input_file1> [<input_file2> ...] <output_file>"
parser = OptionParser(usage=usage, version="%prog - Version 1")
parser.add_option("-b", "--blocksize", action="store", type="int",
dest="blocksize", default="10000000")
options, args = parser.parse_args()
if len(args)==0:
parser.print_help()
sys.exit()
iFilenames = args[:-1]
oFilename = args[-1]
writer = FastaFile(oFilename, 'w', blockSize=10000000)
for iFilename in iFilenames:
for h,s in FastaFile(iFilename):
writer.write(h,s)
writer.close()
#!/usr/bin/env python
"""
exciseRepeats.py
Author: Tony Papenfuss
Date: Fri Mar 21 13:25:51 EST 2008
"""
import os, sys, re
from mungo.fasta import FastaFile
iFilename = sys.argv[1]
oFilename = iFilename + '.excised'
# Cleaned & now masked reads
fi = FastaFile(iFilename)
fo = FastaFile(oFilename, 'w')
for h,s in fi:
s = s.replace('N', '')
fo.write(h,s)
fi.close()
fo.close()
#!/usr/bin/env python
"""
fasta_truncate.py
Author: Tony Papenfuss
Date: July 2013
"""
import sys
import argparse
from mungo.fasta import FastaFile
usage = "%prog <length> <Input file> <Output file>"
parser = argparse.ArgumentParser(description='Truncate fasta files')
parser.add_argument('length', type=int, help='Truncate to length')
parser.add_argument('input_filename', type=str, help='Input filename')
parser.add_argument('output_filename', type=str, help='Output filename')
args = parser.parse_args()
w = FastaFile(args.output_filename, 'w')
for h,s in FastaFile(args.input_filename):
w.write(h, s[0:args.length])
w.close()
"""
fasta2gffAssembly.py <fasta file pattern>
Author: Tony Papenfuss
Date: Mon Feb 25 14:54:57 EST 2008
"""
import os, sys, glob
from mungo.fasta import FastaFile
import mungo.gff as gff
iFilePattern = sys.argv[1]
oFile = open('assembly.gff', 'w')
for filename in glob.iglob(iFilePattern):
f = FastaFile(filename)
for h,L in f.lengthGenerator():
print filename, h, L
g = gff.Feature()
g.reference = h
g.source = 'assembly'
g.type = 'chrom'
g.start = 1
g.end = L
g.group = 'Reference %s' % h
oFile.write(str(g) + '\n')
oFile.flush()
oFile.close()
'--stdin',
dest='stdin',
action='store_true',
help='Input from stdin')
parser.add_option('-o',
'--stdout',
dest='stdout',
action='store_true',
help='Output to stdout')
options, args = parser.parse_args()
if len(args) == 0:
parser.print_help()
sys.exit()
if options.stdin:
f = FastaFile(sys.stdin)
else:
f = FastaFile(args[0])
if options.stdout:
w = FastaFile(sys.stdout, 'w')
else:
w = FastaFile(args[1], 'w')
accList = []
for h, s in f:
acc = h.split()[0]
if not acc in accList:
w.write(h, s)
accList.append(acc)
w.close()
#!/usr/bin/env python
"""
fastaSlice.py <fasta file> <first entry> <number>
Author: Tony Papenfuss
Date: Wed Apr 23 14:49:40 EST 2008
"""
import os, sys
from mungo.fasta import FastaFile
if len(sys.argv)==1 or '-h' in sys.argv:
sys.exit(__doc__)
iFilename = sys.argv[1]
start = int(sys.argv[2])
n = int(sys.argv[3])
f = FastaFile(iFilename, indexed=True)
f.seek(start)
w = FastaFile(sys.stdout, 'w')
count = 0
for h,s in f:
w.write(h,s)
count += 1
if count==n: break
#!/usr/bin/env python
"""
fastaTotalLength.py
Author: Tony Papenfuss
Date: Wed Jun 25 11:13:57 EST 2008
"""
import os, sys
from mungo.fasta import FastaFile
from optparse import OptionParser
usage = "%prog <Input file1> [<Input file2> ...]"
parser = OptionParser(usage=usage, version="%prog - Version 1")
options, args = parser.parse_args()
if len(args) == 0:
parser.print_help()
sys.exit()
total = 0L
for filename in args:
f = FastaFile(filename)
for h, L in f.lengthGenerator():
total += L
print total
#!/usr/bin/env python
"""
fastaOneCharPerLine.py [--stdin] <filename>
Author: Tony Papenfuss
Date: Wed Apr 11 21:41:25 EST 2007
"""
import sys
from optparse import OptionParser
from mungo.fasta import FastaFile
if len(sys.argv) == 1: sys.exit(__doc__)
usage = "%prog [- <] <input fasta file>"
parser = OptionParser(usage=usage, version="%prog - Version 1")
parser.add_option('-s', '--stdin', action='store_true', dest='stdin')
options, args = parser.parse_args()
if options.stdin in args:
faFile = FastaFile(sys.stdin)
else:
faFile = FastaFile(args[0])
for header, seq in faFile:
print '>%s' % header
for b in seq:
print b
print
parser.add_option("-o",
"--output",
dest="oFilename",
help="Output filename",
default=None)
parser.add_option("-w",
"--width",
dest="width",
type="int",
help="Sequence width",
default=60)
options, args = parser.parse_args(sys.argv)
if len(args) != 2: sys.exit(__doc__)
if args[1] != '-':
faFile = FastaFile(args[1])
else:
faFile = FastaFile(sys.stdin)
if options.oFilename:
oFile = open(options.oFilename, 'w')
else:
oFile = sys.stdout
for header, seq in faFile:
protein = sequence.translate(seq)
print >> oFile, '>%s' % header
print >> oFile, pretty(protein, width=options.width)
#!/usr/bin/env python
"""
fastaMaxLength.py
Author: Tony Papenfuss
Date: Wed Jul 8 11:53:34 EST 2009
"""
import os, sys
from mungo.fasta import FastaFile
import numpy
iFilename = sys.argv[1]
maxLength = -1
for h, s in FastaFile(iFilename):
maxLength = max(maxLength, len(s))
print maxLength
#!/usr/bin/env python
"""
exciseRepeats.py
Author: Tony Papenfuss
Date: Fri Mar 21 13:25:51 EST 2008
"""
import os, sys, re
from mungo.fasta import FastaFile
iFilename = sys.argv[1]
oFilename = iFilename + ".excised"
# Cleaned & now masked reads
fi = FastaFile(iFilename)
fo = FastaFile(oFilename, "w")
for h, s in fi:
s = s.replace("N", "")
fo.write(h, s)
fi.close()
fo.close()
#!/usr/bin/env python
"""
fastaHeader.py <fasta file>
Author: Tony Papenfuss
Date: Mon May 7 09:09:27 EST 2007
"""
import os, sys
from mungo.fasta import FastaFile
for h, s in FastaFile(sys.argv[1]):
print h
Author: Tony Papenfuss
Date: Thu Mar 6 11:18:10 EST 2008
"""
import sys
import glob
from mungo.fasta import FastaFile
from optparse import OptionParser
usage = "%prog <Input file1> [<Input file2> ...] <Output file>"
parser = OptionParser(usage=usage, version="%prog - Version 1")
options, args = parser.parse_args()
if len(args) == 0:
parser.print_help()
sys.exit()
iFilenames = args[:-1]
oFilename = args[-1]
writer = FastaFile(oFilename, 'w')
for iFilename in iFilenames:
print >> sys.stderr, '%s-->%s' % (iFilename, oFilename)
f = FastaFile(iFilename)
for h, s in f:
writer.write(h, s)
f.close()
writer.flush()
writer.close()
#!/usr/bin/env python
"""
fasta_average_length.py
Author: Tony Papenfuss
Date: Wed Mar 26 15:03:28 EST 2008
"""
import sys
from mungo.fasta import FastaFile
from optparse import OptionParser
usage = "%prog <Input file1> [<Input file2> ...]"
parser = OptionParser(usage=usage, version="%prog - Version 1")
options, args = parser.parse_args()
if len(args)==0:
parser.print_help()
sys.exit()
for filename in args:
n = 0
average = 0L
for h,s in FastaFile(filename):
n += 1
average += len(s)
print '%s\t%i' % (filename, float(average)/n)
Author: Tony Papenfuss
Date: Thu Mar 6 11:18:10 EST 2008
"""
import sys
import glob
from mungo.fasta import FastaFile
from optparse import OptionParser
usage = "%prog <Input file1> [<Input file2> ...] <Output file>"
parser = OptionParser(usage=usage, version="%prog - Version 1")
options, args = parser.parse_args()
if len(args) == 0:
parser.print_help()
sys.exit()
iFilenames = args[:-1]
oFilename = args[-1]
writer = FastaFile(oFilename, "w")
for iFilename in iFilenames:
print >>sys.stderr, "%s-->%s" % (iFilename, oFilename)
f = FastaFile(iFilename)
for h, s in f:
writer.write(h, s)
f.close()
writer.flush()
writer.close()
"""
fastaOneLine.py [- <] <filename>
Author: Tony Papenfuss
Date: Wed Apr 11 21:41:25 EST 2007
"""
import sys
from optparse import OptionParser
from mungo.fasta import FastaFile
if len(sys.argv)==1: sys.exit(__doc__)
usage = "%prog [- <] <input fasta file>"
parser = OptionParser(usage=usage, version="%prog - Version 1")
options, args = parser.parse_args(sys.argv)
if '-' in args:
iFile = sys.stdin
else:
iFile = open(sys.argv[-1])
for header,seq in FastaFile(iFile):
print '>%s' % header
print seq
print
#!/usr/bin/env python
"""
fasta_truncate.py
Author: Tony Papenfuss
Date: July 2013
"""
import sys
import argparse
from mungo.fasta import FastaFile
usage = "%prog <length> <Input file> <Output file>"
parser = argparse.ArgumentParser(description='Truncate fasta files')
parser.add_argument('length', type=int, help='Truncate to length')
parser.add_argument('input_filename', type=str, help='Input filename')
parser.add_argument('output_filename', type=str, help='Output filename')
args = parser.parse_args()
w = FastaFile(args.output_filename, 'w')
for h, s in FastaFile(args.input_filename):
w.write(h, s[0:args.length])
w.close()
#!/usr/bin/env python
"""
fastaIndex.py <fasta file>
Author: Tony Papenfuss
Date: Wed Apr 23 14:49:40 EST 2008
"""
import os, sys
from mungo.fasta import FastaFile
if len(sys.argv) == 1 or '-h' in sys.argv:
sys.exit(__doc__)
iFilename = sys.argv[1]
f = FastaFile(iFilename, indexed=True)
"""
fastaRename.py <iFilename> <oFilename>
Author: Tony Papenfuss
Date: Fri Mar 7 14:17:14 EST 2008
"""
import sys
from mungo.fasta import FastaFile
iFilename = sys.argv[1]
oFilename = sys.argv[2]
def getSpp(line):
token = line.split('[')[-1].split(']')[0]
return token
writer= FastaFile(oFilename, 'w')
for h,s in FastaFile(iFilename):
tokens = h.split()
name = tokens[0]
spp = getSpp(h)
sppParts = spp.split()
h2 = "%s%s_%s" % (sppParts[0][0:5], sppParts[1][0:2].title(), name)
writer.write(h2, s)
writer.close()