annotate bismark_deduplicate_wrapper.py @ 25:a5faad9e4138 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
author bgruening
date Wed, 22 Aug 2018 08:09:23 -0400
parents
children b2d0e92f81c2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
25
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
1 #!/usr/bin/python
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
2
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
3 import argparse
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
4 import logging
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
5 import os
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
6 import shutil
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
7 import subprocess
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
8 import sys
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
9 import signal
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
10 import tempfile
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
11 from glob import glob
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
12
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
13
a5faad9e4138 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):
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
15 logger.critical(msg)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
16 sys.exit(1)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
17
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
18
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
19 def restore_sigpipe():
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
20 """
a5faad9e4138 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
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
22 """
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
24
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
25
a5faad9e4138 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):
a5faad9e4138 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''):
a5faad9e4138 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())
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
29
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
30
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
31 def get_arg():
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
32 parser = argparse.ArgumentParser()
a5faad9e4138 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')
a5faad9e4138 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')
a5faad9e4138 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')
a5faad9e4138 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')
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
38 args = parser.parse_args()
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
39 return args
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
40
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
41
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
42 def __main__():
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
43 args = get_arg()
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
44
a5faad9e4138 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')
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
46 logger.setLevel(logging.DEBUG)
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
48 if args.log_report:
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
49 ch.setLevel(logging.WARNING)
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
51 handler.setLevel(logging.DEBUG)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
52 logger.addHandler(handler)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
53 else:
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
54 ch.setLevel(logging.DEBUG)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
55 logger.addHandler(ch)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
56
a5faad9e4138 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
a5faad9e4138 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='')
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
59 os.chdir(tmp_dir)
a5faad9e4138 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'
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
62
a5faad9e4138 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'
a5faad9e4138 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']
a5faad9e4138 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))
a5faad9e4138 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,
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
67 preexec_fn=restore_sigpipe)
a5faad9e4138 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()
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
69 logger.info(proc_out)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
70 if process.returncode != 0:
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
72
a5faad9e4138 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'
a5faad9e4138 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'
a5faad9e4138 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)
a5faad9e4138 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 )
a5faad9e4138 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)
a5faad9e4138 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)
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
79 logger.debug("Done.")
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
80
a5faad9e4138 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit 51299fa62f0566a4a897b1c149db564631282fff
bgruening
parents:
diff changeset
81 if __name__=="__main__": __main__()