def main():
# Set defaults
verbose = 0
gapin = "-.~"
gapout = "-"
outfile = ""
Idna = "auto"
Ipep = "auto"
outform = "fasta"
matchmet = "trans"
mtx_file = ""
mtx = None
allinternal = False
readthroughstop = False
# Quick sanity check
if len(sys.argv) < 3:
print __doc__
sys.exit(1)
# Process arguments
dnafile = ""
pepfile = ""
argv = sys.argv[1:]
while (len(argv) > 0):
arg = argv[0]
if arg == "-h":
print __doc__
sys.exit(0)
if arg == "-v": verbose = 1
if arg == "-vv": verbose = 2
if arg == "-vvv": verbose = 3
if arg == "-match":
if len(argv) == 0: argerr("-match")
matchmet = argv[1]
argv = argv[2:]
continue
if arg == "-gapin":
if len(argv) == 0: argerr("-gapin")
gapin = argv[1]
argv = argv[2:]
continue
if arg == "-gapout":
if len(argv) == 0: argerr("-gapout")
gapout = argv[1][0] # Use only the first char
argv = argv[2:]
continue
if arg == "-Idna":
if len(argv) == 0: argerr("-Idna")
Idna = (argv[1]).lower()
argv = argv[2:]
continue
if arg == "-Ipep":
if len(argv) == 0: argerr("-Ipep")
Ipep = (argv[1]).lower()
argv = argv[2:]
continue
if arg == "-O":
if len(argv) == 0: argerr("-O")
outform = (argv[1]).lower()
argv = argv[2:]
continue
if arg == "-mtx":
if len(argv) == 0: argerr("-mtx")
mtx_file = argv[1]
argv = argv[2:]
continue
if arg == "-allinternal":
allinternal = True
if arg == "-readthroughstop":
readthroughstop = True
if arg[0] != "-":
if dnafile == "": dnafile = arg
elif pepfile == "": pepfile = arg
else: outfile = arg
argv = argv[1:]
# Output extra info if requested
if verbose:
warn("verbose level: " + str(verbose))
warn("dnafile: " + dnafile + " [format:" + Idna + "]")
warn("pepfile: " + pepfile + " [format:" + Ipep + "]")
if outfile:
warn("outfile: " + outfile)
else:
warn("outfile: None - writing to STDOUT")
warn("out format: " + outform)
warn("Mtxfile: " + mtx_file)
# Read input files
try:
if Idna == "auto":
Idna = mod_seqfiles.autotype(dnafile)
if verbose: warn("DNA file format appears to be......: " + Idna)
dnaseqs = mod_seqfiles.readfile(dnafile, Idna)
if verbose: warn("#DNA entries read: " + str(len(dnaseqs)))
if Ipep == "auto":
Ipep = mod_seqfiles.autotype(pepfile)
if verbose: warn("Peptide file format appears to be..: " + Ipep)
pepseqs = mod_seqfiles.readfile(pepfile, Ipep)
if verbose: warn("#pep entries read: " + str(len(pepseqs)))
#if 1:
if len(dnaseqs) == 0 or len(pepseqs) == 0:
warn("Error: Bad input.")
warn("Dna sequences read: " + str(len(dnaseqs)))
warn("peptide sequence read: " + str(len(pepseqs)))
sys.exit(1)
if mtx_file:
try:
mtx = mod_translate.parseMatrixFile(mtx_file)
except:
warn("Invalid translation matrix: " + mtx_file)
sys.exit(1)
except Exception, msg:
warn("Error reading input files: " + str(msg))
warn("DNA File type: " + Idna)
warn("Pep File type: " + Ipep)
sys.exit(1)
def main():
# Set defaults
verbose = 0
gapin = "-.~"
gapout = "-"
outfile = ""
Idna = "auto"
Ipep = "auto"
outform = "fasta"
matchmet = "trans"
mtx_file = ""
mtx = None
allinternal = False
readthroughstop = False
# Quick sanity check
if len(sys.argv)<3:
print __doc__
sys.exit(1)
# Process arguments
dnafile = ""
pepfile = ""
argv = sys.argv[1:]
while (len(argv)>0):
arg = argv[0]
if arg == "-h" :
print __doc__
sys.exit(0)
if arg == "-v" : verbose = 1
if arg == "-vv" : verbose = 2
if arg == "-vvv" : verbose = 3
if arg == "-match" :
if len(argv) == 0: argerr("-match")
matchmet = argv[1]
argv = argv[2:]
continue
if arg == "-gapin" :
if len(argv) == 0: argerr("-gapin")
gapin = argv[1]
argv = argv[2:]
continue
if arg == "-gapout" :
if len(argv) == 0: argerr("-gapout")
gapout = argv[1][0] # Use only the first char
argv = argv[2:]
continue
if arg == "-Idna" :
if len(argv) == 0: argerr("-Idna")
Idna = (argv[1]).lower()
argv = argv[2:]
continue
if arg == "-Ipep" :
if len(argv) == 0: argerr("-Ipep")
Ipep = (argv[1]).lower()
argv = argv[2:]
continue
if arg == "-O" :
if len(argv) == 0: argerr("-O")
outform = (argv[1]).lower()
argv = argv[2:]
continue
if arg == "-mtx" :
if len(argv) == 0: argerr("-mtx")
mtx_file = argv[1]
argv = argv[2:]
continue
if arg == "-allinternal":
allinternal = True
if arg == "-readthroughstop":
readthroughstop = True
if arg[0] != "-":
if dnafile == "" : dnafile = arg
elif pepfile == "" : pepfile = arg
else : outfile = arg
argv = argv[1:]
# Output extra info if requested
if verbose:
warn("verbose level: "+str(verbose))
warn("dnafile: "+dnafile+" [format:"+Idna+"]")
warn("pepfile: "+pepfile+" [format:"+Ipep+"]")
if outfile :
warn("outfile: "+outfile)
else :
warn("outfile: None - writing to STDOUT")
warn("out format: "+outform)
warn("Mtxfile: "+mtx_file)
# Read input files
try:
if Idna == "auto":
Idna = mod_seqfiles.autotype(dnafile)
if verbose: warn("DNA file format appears to be......: "+Idna)
dnaseqs = mod_seqfiles.readfile(dnafile,Idna)
if verbose: warn("#DNA entries read: "+str(len(dnaseqs)))
if Ipep == "auto":
Ipep = mod_seqfiles.autotype(pepfile)
if verbose: warn("Peptide file format appears to be..: "+Ipep)
pepseqs = mod_seqfiles.readfile(pepfile,Ipep)
if verbose: warn("#pep entries read: "+str(len(pepseqs)))
#if 1:
if len(dnaseqs) == 0 or len(pepseqs) == 0:
warn("Error: Bad input.")
warn("Dna sequences read: "+str(len(dnaseqs)))
warn("peptide sequence read: "+str(len(pepseqs)))
sys.exit(1)
if mtx_file:
try:
mtx = mod_translate.parseMatrixFile(mtx_file)
except:
warn("Invalid translation matrix: "+mtx_file)
sys.exit(1)
except Exception,msg:
warn("Error reading input files: "+str(msg))
warn("DNA File type: "+Idna)
warn("Pep File type: "+Ipep)
sys.exit(1)
if not opt.outformat in ["AUTO","TAB","FASTA","REPORT"]:
print >> sys.stderr("Invalid output format [%s]\n" % opt.outformat)
sys.exit(-1)
# Check mutually exclusive options
if opt.intronrf: opt.exonann=False
return (opt, args)
if __name__ == "__main__":
reports = []
pepseqs = []
opt, args = parseOpts()
# Initialize translation matrix
mtx = mod_translate.parseMatrixFile(opt.matrix)
if not mtx:
sys.stderr.write("Cannot initialize matrix [%s]\n" % matrix)
sys.exit(-1)
# Start the translation report
reports.append( [REPORTHEADER,"Translation table: %s\n\n" % mtx.description] )
# Read input data
if not args:
sys.stderr.write("Reading from STDIN. -h for help\n")
lines = sys.stdin.readlines()
else:
lines = []
for fn in args:
try:
if not opt.outformat in ["AUTO","TAB","FASTA","REPORT"]:
print >> sys.stderr("Invalid output format [%s]\n" % opt.outformat)
sys.exit(-1)
# Check mutually exclusive options
if opt.intronrf: opt.exonann=False
return (opt, args)
if __name__ == "__main__":
reports = []
pepseqs = []
opt, args = parseOpts()
# Initialize translation matrix
mtx = mod_translate.parseMatrixFile(opt.matrix)
if not mtx:
sys.stderr.write("Cannot initialize matrix [%s]\n" % matrix)
sys.exit(-1)
# Start the translation report
reports.append( [REPORTHEADER,"Translation table: %s\n\n" % mtx.description] )
# Read input data
if not args:
sys.stderr.write("Reading from STDIN. -h for help\n")
lines = sys.stdin.readlines()
else:
lines = []
for fn in args:
try:
if arg == "--readthroughstop":
readThroughStop = True
continue
if arg == "--allinternal":
firstIsStart = False
continue
if arg == "-mtx":
mtxfn = argv[0]
argv = argv[1:]
continue
fn = arg
mtx = mod_translate.parseMatrixFile(mtxfn)
try:
seqs = mod_seqfiles.readfileauto(fn)
newseqs = {}
if not maxseqs: maxseqs = len(seqs.keys())
for key in seqs.keys()[0:maxseqs]:
seq, note = seqs[key]
newseqs[key] = mod_translate.translate(seq, mtx, firstIsStart,
readThroughStop), note
mod_seqfiles.writestream(sys.stdout, newseqs, "fasta", "P")
except Exception, e:
sys.stderr.write("Translation error: %s\n" % str(e))
if arg == "--readthroughstop":
readThroughStop = True
continue
if arg == "--allinternal":
firstIsStart = False
continue
if arg == "-mtx":
mtxfn = argv[0]
argv = argv[1:]
continue
fn = arg
mtx = mod_translate.parseMatrixFile(mtxfn)
try:
seqs = mod_seqfiles.readfileauto(fn)
newseqs = {}
if not maxseqs: maxseqs = len(seqs.keys())
for key in seqs.keys()[0:maxseqs]:
seq, note = seqs[key]
newseqs[key] = mod_translate.translate(seq,mtx,firstIsStart,readThroughStop) , note
mod_seqfiles.writestream(sys.stdout,newseqs,"fasta","P")
except Exception, e:
sys.stderr.write("Translation error: %s\n" % str(e))
sys.exit(1)
