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() f.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()
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() f.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 """ 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 """ 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 """ 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()
""" 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()