Mercurial > repos > peterjc > mira4_assembler

annotate tools/mira4/mira4.py @ 18:381aa262c8cb draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded v0.0.2 preview 10, override /tmp via environment variable
author peterjc
date Tue, 25 Mar 2014 07:37:50 -0400
parents 7fcabeeca5df
children 8487d70e82aa
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
2 """A simple wrapper script to call MIRA and collect its output.
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
3 """
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
4 import os
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
5 import sys
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
6 import subprocess
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
7 import shutil
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
8 import time
18
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
9 import tempfile
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
10
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
11 #Do we need any PYTHONPATH magic?
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
12 from mira4_make_bam import make_bam
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
13
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
14 WRAPPER_VER = "0.0.1" #Keep in sync with the XML file
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
15
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
16 def stop_err(msg, err=1):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
17 sys.stderr.write(msg+"\n")
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
18 sys.exit(err)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
19
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
20
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
21 def get_version(mira_binary):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
22 """Run MIRA to find its version number"""
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
23 # At the commend line I would use: mira -v | head -n 1
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
24 # however there is some pipe error when doing that here.
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
25 cmd = [mira_binary, "-v"]
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
26 try:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
27 child = subprocess.Popen(cmd,
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
28 stdout=subprocess.PIPE,
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
29 stderr=subprocess.STDOUT)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
30 except Exception, err:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
31 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
32 sys.exit(1)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
33 ver, tmp = child.communicate()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
34 del child
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
35 return ver.split("\n", 1)[0].strip()
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
36
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
37
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
38 try:
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
39 mira_path = os.environ["MIRA4"]
12
a6a56440567c Uploaded v0.0.2 preview 4, now using MIRA 4.0 (not RC), and sets MIRA4 environment variable during install.
peterjc
parents: 11
diff changeset
40 except KeyError:
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
41 stop_err("Environment variable $MIRA4 not set")
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
42 mira_binary = os.path.join(mira_path, "mira")
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
43 if not os.path.isfile(mira_binary):
13
7fcabeeca5df Uploaded v0.0.2 preview 5, fixes for MIRA 4.0 (final), more verbose error if $MIRA4 path wrong
peterjc
parents: 12
diff changeset
44 stop_err("Missing mira under $MIRA4, %r\nFolder contained: %s"
7fcabeeca5df Uploaded v0.0.2 preview 5, fixes for MIRA 4.0 (final), more verbose error if $MIRA4 path wrong
peterjc
parents: 12
diff changeset
45 % (mira_binary, ", ".join(os.listdir(mira_path))))
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
46 mira_convert = os.path.join(mira_path, "miraconvert")
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
47 if not os.path.isfile(mira_convert):
13
7fcabeeca5df Uploaded v0.0.2 preview 5, fixes for MIRA 4.0 (final), more verbose error if $MIRA4 path wrong
peterjc
parents: 12
diff changeset
48 stop_err("Missing miraconvert under $MIRA4, %r\nFolder contained: %s"
7fcabeeca5df Uploaded v0.0.2 preview 5, fixes for MIRA 4.0 (final), more verbose error if $MIRA4 path wrong
peterjc
parents: 12
diff changeset
49 % (mira_convert, ", ".join(os.listdir(mira_path))))
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
50
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
51 mira_ver = get_version(mira_binary)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
52 if not mira_ver.strip().startswith("4.0"):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
53 stop_err("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_binary))
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
54 mira_convert_ver = get_version(mira_convert)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
55 if not mira_convert_ver.strip().startswith("4.0"):
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
56 stop_err("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_convert))
5
ffefb87bd414 Uploaded v0.0.1 preview 5, using MIRA 4.0 RC4, supports segment_placement (pairing type)
peterjc
parents: 4
diff changeset
57 if "-v" in sys.argv or "--version" in sys.argv:
ffefb87bd414 Uploaded v0.0.1 preview 5, using MIRA 4.0 RC4, supports segment_placement (pairing type)
peterjc
parents: 4
diff changeset
58 print "%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
59 if mira_ver != mira_convert_ver:
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
60 print "WARNING: miraconvert %s" % mira_convert_ver
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
61 sys.exit(0)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
62
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
63
11
02350bef2e99 Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents: 9
diff changeset
64 try:
02350bef2e99 Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents: 9
diff changeset
65 threads = int(os.environ.get("GALAXY_SLOTS", "1"))
02350bef2e99 Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents: 9
diff changeset
66 except ValueError:
02350bef2e99 Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents: 9
diff changeset
67 threads = 1
02350bef2e99 Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents: 9
diff changeset
68 assert 1 <= threads, threads
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
69
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
70
18
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
71 def override_temp(manifest):
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
72 """Override ``-DI:trt=/tmp`` in manifest with environment variable.
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
73
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
74 Currently MIRA 4 does not allow envronment variables like ``$TMP``
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
75 inside the manifest, which is a problem if you need to override
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
76 the default at run time.
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
77
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
78 The tool XML will ``/tmp`` and we replace that here with
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
79 ``tempfile.gettempdir()`` which will respect $TMPDIR, $TEMP, $TMP
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
80 as explained in the Python standard library documentation:
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
81 http://docs.python.org/2/library/tempfile.html#tempfile.tempdir
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
82
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
83 By default MIRA 4 would write its temporary files within the output
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
84 folder, which is a problem if that is a network drive.
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
85 """
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
86 handle = open(manifest, "r")
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
87 text = handle.read()
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
88 handle.close()
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
89
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
90 #At time of writing, this is at the end of a file,
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
91 #but could be followed by a space in future...
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
92 text = text.replace("-DI:trt=/tmp", "-DI:trt=" + tempfile.gettempdir())
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
93
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
94 handle = open(manifest, "w")
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
95 handle.write(text)
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
96 handle.flush()
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
97 handle.close()
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
98
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
99
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
100 def log_manifest(manifest):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
101 """Write the manifest file to stderr."""
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
102 sys.stderr.write("\n%s\nManifest file\n%s\n" % ("="*60, "="*60))
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
103 with open(manifest) as h:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
104 for line in h:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
105 sys.stderr.write(line)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
106 sys.stderr.write("\n%s\nEnd of manifest\n%s\n" % ("="*60, "="*60))
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
107
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
108
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
109 def collect_output(temp, name, handle):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
110 """Moves files to the output filenames (global variables)."""
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
111 n3 = (temp, name, name, name)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
112 f = "%s/%s_assembly/%s_d_results" % (temp, name, name)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
113 if not os.path.isdir(f):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
114 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
115 stop_err("Missing output folder")
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
116 if not os.listdir(f):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
117 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
118 stop_err("Empty output folder")
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
119 missing = []
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
120
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
121 old_maf = "%s/%s_out.maf" % (f, name)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
122 if not os.path.isfile(old_maf):
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
123 #Triggered extractLargeContigs.sh?
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
124 old_maf = "%s/%s_LargeContigs_out.maf" % (f, name)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
125
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
126 #De novo or single strain mapping,
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
127 old_fasta = "%s/%s_out.unpadded.fasta" % (f, name)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
128 ref_fasta = "%s/%s_out.padded.fasta" % (f, name)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
129 if not os.path.isfile(old_fasta):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
130 #Mapping (StrainX versus reference) or de novo
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
131 old_fasta = "%s/%s_out_StrainX.unpadded.fasta" % (f, name)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
132 ref_fasta = "%s/%s_out_StrainX.padded.fasta" % (f, name)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
133 if not os.path.isfile(old_fasta):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
134 old_fasta = "%s/%s_out_ReferenceStrain.unpadded.fasta" % (f, name)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
135 ref_fasta = "%s/%s_out_ReferenceStrain.padded.fasta" % (f, name)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
136
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
137
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
138 missing = False
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
139 for old, new in [(old_maf, out_maf),
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
140 (old_fasta, out_fasta)]:
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
141 if not os.path.isfile(old):
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
142 missing = True
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
143 else:
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
144 handle.write("Capturing %s\n" % old)
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
145 shutil.move(old, new)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
146 if missing:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
147 log_manifest(manifest)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
148 sys.stderr.write("Contents of %r:\n" % f)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
149 for filename in sorted(os.listdir(f)):
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
150 sys.stderr.write("%s\n" % filename)
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
151
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
152 #For mapping mode, probably most people would expect a BAM file
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
153 #using the reference FASTA file...
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
154 msg = make_bam(mira_convert, out_maf, ref_fasta, out_bam, handle)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
155 if msg:
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
156 stop_err(msg)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
157
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
158 def clean_up(temp, name):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
159 folder = "%s/%s_assembly" % (temp, name)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
160 if os.path.isdir(folder):
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
161 shutil.rmtree(folder)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
162
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
163 #TODO - Run MIRA in /tmp or a configurable directory?
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
164 #Currently Galaxy puts us somewhere safe like:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
165 #/opt/galaxy-dist/database/job_working_directory/846/
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
166 temp = "."
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
167 #name, out_fasta, out_qual, out_ace, out_caf, out_wig, out_log = sys.argv[1:8]
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
168 name = "MIRA"
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
169 manifest, out_maf, out_bam, out_fasta, out_log = sys.argv[1:]
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
170
18
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
171 override_temp(manifest)
381aa262c8cb Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents: 13
diff changeset
172
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
173 start_time = time.time()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
174 #cmd_list =sys.argv[8:]
11
02350bef2e99 Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents: 9
diff changeset
175 cmd_list = [mira_binary, "-t", str(threads), manifest]
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
176 cmd = " ".join(cmd_list)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
177
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
178 assert os.path.isdir(temp)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
179 d = "%s_assembly" % name
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
180 assert not os.path.isdir(d), "Path %s already exists" % d
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
181 try:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
182 #Check path access
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
183 os.mkdir(d)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
184 except Exception, err:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
185 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
186 sys.stderr.write("Error making directory %s\n%s" % (d, err))
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
187 sys.exit(1)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
188
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
189 #print os.path.abspath(".")
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
190 #print cmd
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
191
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
192 handle = open(out_log, "w")
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
193 handle.write("======================== MIRA manifest (instructions) ========================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
194 m = open(manifest, "rU")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
195 for line in m:
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
196 handle.write(line)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
197 m.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
198 del m
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
199 handle.write("\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
200 handle.write("============================ Starting MIRA now ===============================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
201 handle.flush()
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
202 try:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
203 #Run MIRA
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
204 child = subprocess.Popen(cmd_list,
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
205 stdout=handle,
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
206 stderr=subprocess.STDOUT)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
207 except Exception, err:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
208 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
209 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
210 #TODO - call clean up?
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
211 handle.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
212 handle.close()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
213 sys.exit(1)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
214 #Use .communicate as can get deadlocks with .wait(),
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
215 stdout, stderr = child.communicate()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
216 assert not stdout and not stderr #Should be empty as sent to handle
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
217 run_time = time.time() - start_time
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
218 return_code = child.returncode
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
219 handle.write("\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
220 handle.write("============================ MIRA has finished ===============================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
221 handle.write("MIRA took %0.2f hours\n" % (run_time / 3600.0))
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
222 if return_code:
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
223 print "MIRA took %0.2f hours" % (run_time / 3600.0)
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
224 handle.write("Return error code %i from command:\n" % return_code)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
225 handle.write(cmd + "\n")
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
226 handle.close()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
227 clean_up(temp, name)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
228 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
229 stop_err("Return error code %i from command:\n%s" % (return_code, cmd),
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
230 return_code)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
231 handle.flush()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
232
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
233 if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"):
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
234 handle.write("\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
235 handle.write("====================== Extract Large Contigs failed ==========================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
236 e = open("MIRA_assembly/MIRA_d_results/ec.log", "rU")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
237 for line in e:
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
238 handle.write(line)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
239 e.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
240 handle.write("============================ (end of ec.log) =================================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
241 handle.flush()
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
242
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
243 #print "Collecting output..."
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
244 start_time = time.time()
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
245 collect_output(temp, name, handle)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
246 collect_time = time.time() - start_time
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
247 handle.write("MIRA took %0.2f hours; collecting output %0.2f minutes\n" % (run_time / 3600.0, collect_time / 60.0))
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
248 print("MIRA took %0.2f hours; collecting output %0.2f minutes\n" % (run_time / 3600.0, collect_time / 60.0))
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
249
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
250 if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"):
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
251 #Treat as an error, but doing this AFTER collect_output
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
252 sys.stderr.write("Extract Large Contigs failed\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
253 handle.write("Extract Large Contigs failed\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
254 handle.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
255 sys.exit(1)
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
256
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
257 #print "Cleaning up..."
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
258 clean_up(temp, name)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
259
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
260 handle.write("\nDone\n")
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
261 handle.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
262 print("Done")