comparison tools/get_orfs_or_cdss/get_orfs_or_cdss.py @ 11:d51db443aaa4 draft

v0.2.3 Python 3 compatible print function.

10:01b0488d9cfb

(formerly SCRI), Dundee, UK. All rights reserved.

See accompanying text file for licence details (MIT licence).
"""

import re
import sys

from optparse import OptionParser

usage = """Use as follows:

$ python get_orfs_or_cdss.py -i genome.fa -f fasta --table 11 -t CDS -e open -m all -s both --on cds.nuc.fa --op cds.protein.fa --ob cds.bed --og cds.gff3
"""

try:
    from Bio.Seq import Seq, reverse_complement, translate
    from Bio.SeqRecord import SeqRecord

                  help='Show version and quit')

options, args = parser.parse_args()

if options.version:
    print("v0.2.0")
    sys.exit(0)

if not options.input_file:
    sys.exit("Input file is required")

elif options.seq_format.lower().startswith("fastq"):
    seq_format = "fastq"
else:
    sys.exit("Unsupported file type %r" % options.seq_format)

print "Genetic code table %i" % options.table
print "Minimum length %i aa" % options.min_len
# print "Taking %s ORF(s) from %s strand(s)" % (mode, strand)

starts = sorted(table_obj.start_codons)
assert "NNN" not in starts
re_starts = re.compile("|".join(starts))

if out_prot and out_prot is not sys.stdout:
    out_prot.close()
if out_bed and out_bed is not sys.stdout:
    out_bed.close()

print "Found %i %ss in %i sequences" % (out_count, options.ftype, in_count)

11:d51db443aaa4

(formerly SCRI), Dundee, UK. All rights reserved.

See accompanying text file for licence details (MIT licence).
"""

from __future__ import print_function

import re
import sys

from optparse import OptionParser

usage = r"""Use as follows:

$ python get_orfs_or_cdss.py -i genome.fa -f fasta --table 11 \
-t CDS -e open -m all -s both --on cds.nuc.fa --op cds.protein.fa \
--ob cds.bed --og cds.gff3
"""

try:
    from Bio.Seq import Seq, reverse_complement, translate
    from Bio.SeqRecord import SeqRecord

                  help='Show version and quit')

options, args = parser.parse_args()

if options.version:
    print("v0.2.3")
    sys.exit(0)

if not options.input_file:
    sys.exit("Input file is required")

elif options.seq_format.lower().startswith("fastq"):
    seq_format = "fastq"
else:
    sys.exit("Unsupported file type %r" % options.seq_format)

print("Genetic code table %i" % options.table)
print("Minimum length %i aa" % options.min_len)
# print "Taking %s ORF(s) from %s strand(s)" % (mode, strand)

starts = sorted(table_obj.start_codons)
assert "NNN" not in starts
re_starts = re.compile("|".join(starts))

if out_prot and out_prot is not sys.stdout:
    out_prot.close()
if out_bed and out_bed is not sys.stdout:
    out_bed.close()

print("Found %i %ss in %i sequences" % (out_count, options.ftype, in_count))