Mercurial > repos > peterjc > samtools_idxstats

annotate tools/samtools_idxstats/samtools_idxstats.py @ 0:d4412c04d7b1 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
author peterjc
date Wed, 20 Nov 2013 12:27:33 -0500
parents
children 8945bad80f4a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
2 """Wrapper for "samtools idxstats" for use in Galaxy.
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
3
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
4 This script takes exactly three command line arguments:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
5 * Input BAM filename
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
6 * Input BAI filename (via Galaxy metadata)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
7 * Output tabular filename
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
8
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
9 This messes about with the filenames to make samtools happy, then
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
10 runs "samtools idxstats" and captures the output to the desired
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
11 tabular file.
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
12 """
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
13 import sys
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
14 import os
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
15 import subprocess
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
16 import tempfile
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
17
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
18 if "-v" in sys.argv or "--version" in sys.argv:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
19 #Galaxy seems to invert the order of the two lines
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
20 print "(Galaxy wrapper v0.0.1)"
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
21 cmd = "samtools 2>&1 | grep -i ^Version"
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
22 sys.exit(os.system(cmd))
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
23
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
24 def stop_err(msg, error_level=1):
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
25 """Print error message to stdout and quit with given error level."""
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
26 sys.stderr.write("%s\n" % msg)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
27 sys.exit(error_level)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
28
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
29 if len(sys.argv) != 4:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
30 stop_err("Require three arguments: BAM, BAI, tabular filenames")
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
31
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
32 bam_filename, bai_filename, tabular_filename = sys.argv[1:]
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
33
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
34 if not os.path.isfile(bam_filename):
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
35 stop_err("Input BAM file not found: %s" % bam_filename)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
36 if not os.path.isfile(bai_filename):
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
37 if bai_filename == "None":
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
38 stop_err("Error: Galaxy did not index your BAM file")
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
39 stop_err("Input BAI file not found: %s" % bai_filename)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
40
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
41 #Assign sensible names with real extensions, and setup symlinks:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
42 tmp_dir = tempfile.mkdtemp()
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
43 bam_file = os.path.join(tmp_dir, "temp.bam")
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
44 bai_file = os.path.join(tmp_dir, "temp.bam.bai")
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
45 os.symlink(os.path.abspath(bam_filename), bam_file)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
46 os.symlink(os.path.abspath(bai_filename), bai_file)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
47 assert os.path.isfile(bam_file), bam_file
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
48 assert os.path.isfile(bai_file), bai_file
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
49 assert os.path.isfile(bam_file + ".bai"), bam_file
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
50
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
51 #Run samtools idxstats:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
52 cmd = "samtools idxstats %s > %s" % (bam_file, tabular_filename)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
53 return_code = os.system(cmd)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
54
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
55 #Remove the temp symlinks:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
56 os.remove(bam_file)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
57 os.remove(bai_file)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
58 os.rmdir(tmp_dir)
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
59
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
60 if return_code:
d4412c04d7b1 Uploaded v0.0.1 (as tested previously on the Test Tool Shed)
peterjc
parents:
diff changeset
61 stop_err("Return code %i from command:\n%s" % (return_code, cmd))