def main():
# Read options, args.
parser = optparse.OptionParser()
parser.add_option('-c', '--chr-col', type='int', dest='chrom_col')
parser.add_option('-s', '--start-col', type='int', dest='start_col')
parser.add_option('-e', '--end-col', type='int', dest='end_col')
parser.add_option('-P', '--preset', dest='preset')
(options, args) = parser.parse_args()
input_fname, output_fname = args
tmpfile = tempfile.NamedTemporaryFile()
sort_params = None
if options.chrom_col and options.start_col and options.end_col:
sort_params = [
"sort",
"-k%(i)s,%(i)s" % {'i': options.chrom_col},
"-k%(i)i,%(i)in" % {'i': options.start_col},
"-k%(i)i,%(i)in" % {'i': options.end_col}
]
elif options.preset == "bed":
sort_params = ["sort", "-k1,1", "-k2,2n", "-k3,3n"]
elif options.preset == "vcf":
sort_params = ["sort", "-k1,1", "-k2,2n"]
elif options.preset == "gff":
sort_params = ["sort", "-s", "-k1,1", "-k4,4n"] # stable sort on start column
# Skip any lines starting with "#" and "track"
grepped = subprocess.Popen(["grep", "-e", "^\"#\"", "-e", "^track", "-v", input_fname], stderr=subprocess.PIPE, stdout=subprocess.PIPE)
after_sort = subprocess.Popen(sort_params, stdin=grepped.stdout, stderr=subprocess.PIPE, stdout=tmpfile)
grepped.stdout.close()
output, err = after_sort.communicate()
ctabix.tabix_compress(tmpfile.name, output_fname, force=True)
def main():
# Read options, args.
parser = optparse.OptionParser()
(options, args) = parser.parse_args()
input_fname, output_fname = args
ctabix.tabix_compress(input_fname, output_fname, force=True)
def main():
# Read options, args.
usage = "Usage: %prog [options] tabular_input_file bgzip_output_file"
parser = optparse.OptionParser(usage=usage)
parser.add_option('-c',
'--chr-col',
type='int',
default=0,
dest='chrom_col')
parser.add_option('-s',
'--start-col',
type='int',
default=1,
dest='start_col')
parser.add_option('-e', '--end-col', type='int', default=1, dest='end_col')
(options, args) = parser.parse_args()
if len(args) != 2:
parser.print_usage()
exit(1)
input_fname, output_fname = args
output_dir = os.path.dirname(output_fname)
if not os.path.exists(output_dir):
os.makedirs(output_dir)
ctabix.tabix_compress(input_fname, output_fname, force=True)
# Column indices are 0-based.
ctabix.tabix_index(output_fname,
seq_col=options.chrom_col,
start_col=options.start_col,
end_col=options.end_col)
def main():
# Read options, args.
parser = optparse.OptionParser()
parser.add_option('-c', '--chr-col', type='int', dest='chrom_col')
parser.add_option('-s', '--start-col', type='int', dest='start_col')
parser.add_option('-e', '--end-col', type='int', dest='end_col')
parser.add_option('-P', '--preset', dest='preset')
(options, args) = parser.parse_args()
input_fname, output_fname = args
tmpfile = tempfile.NamedTemporaryFile()
sort_params = None
if options.chrom_col and options.start_col and options.end_col:
sort_params = [
"sort",
"-k%(i)s,%(i)s" % {
'i': options.chrom_col
},
"-k%(i)i,%(i)in" % {
'i': options.start_col
},
"-k%(i)i,%(i)in" % {
'i': options.end_col
}
]
elif options.preset == "bed":
sort_params = ["sort", "-k1,1", "-k2,2n", "-k3,3n"]
elif options.preset == "vcf":
sort_params = ["sort", "-k1,1", "-k2,2n"]
elif options.preset == "gff":
sort_params = ["sort", "-s", "-k1,1",
"-k4,4n"] # stable sort on start column
# Skip any lines starting with "#" and "track"
grepped = subprocess.Popen(
["grep", "-e", "^\"#\"", "-e", "^track", "-v", input_fname],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
after_sort = subprocess.Popen(sort_params,
stdin=grepped.stdout,
stderr=subprocess.PIPE,
stdout=tmpfile)
grepped.stdout.close()
output, err = after_sort.communicate()
ctabix.tabix_compress(tmpfile.name, output_fname, force=True)
def main():
# Read options, args.
usage = "Usage: %prog [options] tabular_input_file bgzip_output_file"
parser = optparse.OptionParser(usage=usage)
parser.add_option( '-c', '--chr-col', type='int', default=0, dest='chrom_col' )
parser.add_option( '-s', '--start-col', type='int', default=1, dest='start_col' )
parser.add_option( '-e', '--end-col', type='int', default=1, dest='end_col' )
(options, args) = parser.parse_args()
if len(args) != 2:
parser.print_usage()
exit(1)
input_fname, output_fname = args
output_dir = os.path.dirname(output_fname)
if not os.path.exists(output_dir):
os.makedirs(output_dir)
ctabix.tabix_compress(input_fname, output_fname, force=True)
# Column indices are 0-based.
ctabix.tabix_index(output_fname, seq_col=options.chrom_col, start_col=options.start_col, end_col=options.end_col)
def tabix_file(input_fname, output_fname):
print >> sys.stdout, "tabix_file: %s -> %s" % (input_fname, output_fname)
ctabix.tabix_compress(input_fname, output_fname, force=True)
# Column indices are 0-based.
ctabix.tabix_index(output_fname, seq_col=0, start_col=1, end_col=1)
def tabix_file(input_fname, output_fname):
print >> sys.stdout, "tabix_file: %s -> %s" % (input_fname, output_fname)
ctabix.tabix_compress(input_fname, output_fname, force=True)
# Column indices are 0-based.
ctabix.tabix_index(output_fname, seq_col=0, start_col=1, end_col=1)
