Mercurial > repos > bgruening > bismark

annotate bismark_deduplicate_wrapper.py @ 20:ff6ee551b153 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 2d8f9bfd501f1cd82b5cf0429c1b02b3784392ab"
author bgruening
date Fri, 04 Oct 2019 11:33:27 -0400
parents 9bfe38410155
children 120b7b35e442
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
8
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
1 #!/usr/bin/python
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
2
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
3 import argparse
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
4 import logging
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
5 import os
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
6 import shutil
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
7 import subprocess
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
8 import sys
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
9 import signal
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
10 import tempfile
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
11 from glob import glob
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
12
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
13
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
14 def stop_err(logger, msg):
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
15 logger.critical(msg)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
16 sys.exit(1)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
17
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
18
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
19 def restore_sigpipe():
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
20 """
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
21 Needed to handle samtools view
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
22 """
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
23 signal.signal(signal.SIGPIPE, signal.SIG_DFL)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
24
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
25
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
26 def log_subprocess_output(logger, pipe):
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
27 for line in iter(pipe.readline, b''):
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
28 logger.debug(line.decode().rstrip())
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
29
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
30
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
31 def get_arg():
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
32 parser = argparse.ArgumentParser()
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
33 parser.add_argument('--single_or_paired', dest='single_or_paired')
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
34 parser.add_argument('--input', dest='input', metavar='input')
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
35 parser.add_argument('--output_report', dest='output_report', metavar='output_report')
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
36 parser.add_argument('--output_bam', dest='output_bam', metavar='output_report')
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
37 parser.add_argument('--log_report', dest='log_report', metavar='log_filename', type=str)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
38 args = parser.parse_args()
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
39 return args
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
40
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
41
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
42 def __main__():
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
43 args = get_arg()
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
44
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
45 logger = logging.getLogger('bismark_deduplicate_wrapper')
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
46 logger.setLevel(logging.DEBUG)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
47 ch = logging.StreamHandler(sys.stdout)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
48 if args.log_report:
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
49 ch.setLevel(logging.WARNING)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
50 handler = logging.FileHandler(args.log_report)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
51 handler.setLevel(logging.DEBUG)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
52 logger.addHandler(handler)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
53 else:
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
54 ch.setLevel(logging.DEBUG)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
55 logger.addHandler(ch)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
56
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
57 # ensure the input has a .bam suffix
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
58 tmp_dir = tempfile.mkdtemp(prefix='tmp', suffix='')
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
59 os.chdir(tmp_dir)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
60 default_reads_name = 'submitted_reads.bam'
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
61 os.symlink(args.input, default_reads_name)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
62
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
63 single_or_paired = '-s' if args.single_or_paired == 'single' else '-p'
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
64 cmd = ['deduplicate_bismark', single_or_paired, default_reads_name, '--bam']
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
65 logger.info("Deduplicating with: '%s'", " ".join(cmd))
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
66 process = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT,
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
67 preexec_fn=restore_sigpipe)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
68 proc_out, proc_err = process.communicate()
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
69 logger.info(proc_out)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
70 if process.returncode != 0:
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
71 stop_err(logger, "Bismark deduplication error (also check the log file if any)!\n%s" % proc_err)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
72
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
73 deduplicated_out_name = 'submitted_reads.deduplicated.bam'
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
74 deduplicated_report_name = 'submitted_reads.deduplication_report.txt'
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
75 logger.debug("Moving '%s' to galaxy: '%s'.", deduplicated_out_name, args.output_bam)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
76 shutil.move(deduplicated_out_name, args.output_bam )
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
77 logger.debug("Moving '%s' to galaxy: '%s'.", deduplicated_report_name, args.output_report)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
78 shutil.move('submitted_reads.deduplication_report.txt', args.output_report)
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
79 logger.debug("Done.")
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
80
9bfe38410155 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
81 if __name__=="__main__": __main__()