Mercurial > repos > peterjc > mira4_assembler
annotate tools/mira4/mira4.py @ 21:4abe8d59a438 draft
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
author | peterjc |
---|---|
date | Tue, 28 Oct 2014 08:29:59 -0400 |
parents | aeb3e35f8236 |
children |
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 |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
10 from optparse import OptionParser |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
11 |
9 | 12 #Do we need any PYTHONPATH magic? |
13 from mira4_make_bam import make_bam | |
14 | |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
15 WRAPPER_VER = "0.0.4" #Keep in sync with the XML file |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
16 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
17 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
|
18 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
|
19 sys.exit(err) |
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 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
22 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
|
23 """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
|
24 # 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
|
25 # 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
|
26 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
|
27 try: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
28 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
|
29 stdout=subprocess.PIPE, |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
30 stderr=subprocess.STDOUT) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
31 except Exception, err: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
32 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
|
33 sys.exit(1) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
34 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
|
35 del child |
9 | 36 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
|
37 |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
38 #Parse Command Line |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
39 usage = """Galaxy MIRA4 wrapper script v%s - use as follows: |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
40 |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
41 $ python mira4.py ... |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
42 |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
43 This will run the MIRA binary and collect its output files as directed. |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
44 """ % WRAPPER_VER |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
45 parser = OptionParser(usage=usage) |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
46 parser.add_option("-m", "--manifest", dest="manifest", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
47 default=None, metavar="FILE", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
48 help="MIRA manifest filename") |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
49 parser.add_option("--maf", dest="maf", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
50 default="-", metavar="FILE", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
51 help="MIRA MAF output filename") |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
52 parser.add_option("--bam", dest="bam", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
53 default="-", metavar="FILE", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
54 help="Unpadded BAM output filename") |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
55 parser.add_option("--fasta", dest="fasta", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
56 default="-", metavar="FILE", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
57 help="Unpadded FASTA output filename") |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
58 parser.add_option("--log", dest="log", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
59 default="-", metavar="FILE", |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
60 help="MIRA logging output filename") |
21
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
61 parser.add_option("-v", "--version", dest="version", |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
62 default=False, action="store_true", |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
63 help="Show version and quit") |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
64 options, args = parser.parse_args() |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
65 manifest = options.manifest |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
66 out_maf = options.maf |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
67 out_bam = options.bam |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
68 out_fasta = options.fasta |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
69 out_log = options.log |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
70 |
7 | 71 try: |
72 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
|
73 except KeyError: |
7 | 74 stop_err("Environment variable $MIRA4 not set") |
75 mira_binary = os.path.join(mira_path, "mira") | |
76 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
|
77 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
|
78 % (mira_binary, ", ".join(os.listdir(mira_path)))) |
9 | 79 mira_convert = os.path.join(mira_path, "miraconvert") |
80 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
|
81 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
|
82 % (mira_convert, ", ".join(os.listdir(mira_path)))) |
7 | 83 |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
84 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
|
85 if not mira_ver.strip().startswith("4.0"): |
9 | 86 stop_err("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_binary)) |
87 mira_convert_ver = get_version(mira_convert) | |
88 if not mira_convert_ver.strip().startswith("4.0"): | |
89 stop_err("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_convert)) | |
21
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
90 if options.version: |
5
ffefb87bd414
Uploaded v0.0.1 preview 5, using MIRA 4.0 RC4, supports segment_placement (pairing type)
peterjc
parents:
4
diff
changeset
|
91 print "%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER) |
9 | 92 if mira_ver != mira_convert_ver: |
93 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
|
94 sys.exit(0) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
95 |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
96 if not manifest: |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
97 stop_err("Manifest is required") |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
98 elif not os.path.isfile(manifest): |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
99 stop_err("Missing input MIRA manifest file: %r" % manifest) |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
100 |
7 | 101 |
11
02350bef2e99
Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents:
9
diff
changeset
|
102 try: |
02350bef2e99
Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents:
9
diff
changeset
|
103 threads = int(os.environ.get("GALAXY_SLOTS", "1")) |
02350bef2e99
Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents:
9
diff
changeset
|
104 except ValueError: |
02350bef2e99
Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents:
9
diff
changeset
|
105 threads = 1 |
02350bef2e99
Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents:
9
diff
changeset
|
106 assert 1 <= threads, threads |
7 | 107 |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
108 |
18
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
109 def override_temp(manifest): |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
110 """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
|
111 |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
112 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
|
113 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
|
114 the default at run time. |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
115 |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
116 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
|
117 ``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
|
118 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
|
119 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
|
120 |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
121 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
|
122 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
|
123 """ |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
124 handle = open(manifest, "r") |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
125 text = handle.read() |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
126 handle.close() |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
127 |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
128 #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
|
129 #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
|
130 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
|
131 |
19 | 132 #Want to try to ensure this gets written to disk before MIRA attempts |
133 #to open it - any networked file system may impose a delay... | |
18
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
134 handle = open(manifest, "w") |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
135 handle.write(text) |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
136 handle.flush() |
19 | 137 os.fsync(handle.fileno()) |
18
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
138 handle.close() |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
139 |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
140 |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
141 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
|
142 """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
|
143 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
|
144 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
|
145 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
|
146 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
|
147 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
|
148 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
149 |
4 | 150 def collect_output(temp, name, handle): |
9 | 151 """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
|
152 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
|
153 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
|
154 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
|
155 log_manifest(manifest) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
156 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
|
157 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
|
158 log_manifest(manifest) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
159 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
|
160 missing = [] |
4 | 161 |
162 old_maf = "%s/%s_out.maf" % (f, name) | |
163 if not os.path.isfile(old_maf): | |
164 #Triggered extractLargeContigs.sh? | |
165 old_maf = "%s/%s_LargeContigs_out.maf" % (f, name) | |
166 | |
167 #De novo or single strain mapping, | |
168 old_fasta = "%s/%s_out.unpadded.fasta" % (f, name) | |
9 | 169 ref_fasta = "%s/%s_out.padded.fasta" % (f, name) |
4 | 170 if not os.path.isfile(old_fasta): |
9 | 171 #Mapping (StrainX versus reference) or de novo |
4 | 172 old_fasta = "%s/%s_out_StrainX.unpadded.fasta" % (f, name) |
9 | 173 ref_fasta = "%s/%s_out_StrainX.padded.fasta" % (f, name) |
4 | 174 if not os.path.isfile(old_fasta): |
9 | 175 old_fasta = "%s/%s_out_ReferenceStrain.unpadded.fasta" % (f, name) |
176 ref_fasta = "%s/%s_out_ReferenceStrain.padded.fasta" % (f, name) | |
177 | |
4 | 178 |
179 missing = False | |
180 for old, new in [(old_maf, out_maf), | |
181 (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
|
182 if not os.path.isfile(old): |
4 | 183 missing = True |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
184 elif not new or new == "-": |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
185 handle.write("Ignoring %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
|
186 else: |
4 | 187 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
|
188 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
|
189 if missing: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
190 log_manifest(manifest) |
4 | 191 sys.stderr.write("Contents of %r:\n" % f) |
192 for filename in sorted(os.listdir(f)): | |
193 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
|
194 |
9 | 195 #For mapping mode, probably most people would expect a BAM file |
196 #using the reference FASTA file... | |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
197 if out_bam and out_bam != "-": |
21
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
198 if out_maf and out_maf != "-": |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
199 msg = make_bam(mira_convert, out_maf, ref_fasta, out_bam, handle) |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
200 else: |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
201 #Not collecting the MAF file, use original location |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
202 msg = make_bam(mira_convert, old_maf, ref_fasta, out_bam, handle) |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
203 if msg: |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
204 stop_err(msg) |
9 | 205 |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
206 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
|
207 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
|
208 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
|
209 shutil.rmtree(folder) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
210 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
211 #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
|
212 #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
|
213 #/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
|
214 temp = "." |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
215 |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
216 name = "MIRA" |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
217 |
18
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
218 override_temp(manifest) |
381aa262c8cb
Uploaded v0.0.2 preview 10, override /tmp via environment variable
peterjc
parents:
13
diff
changeset
|
219 |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
220 start_time = time.time() |
11
02350bef2e99
Uploaded v0.0.2 preview 3, correction regression in URL
peterjc
parents:
9
diff
changeset
|
221 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
|
222 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
|
223 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
224 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
|
225 d = "%s_assembly" % name |
21
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
226 #This can fail on my development machine if stale folders exist |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
227 #under Galaxy's .../database/job_working_directory/ tree: |
4abe8d59a438
Uploaded v0.0.4 preview 1; fix getting BAM without MAF
peterjc
parents:
20
diff
changeset
|
228 assert not os.path.isdir(d), "Path %r already exists:\n%s" % (d, os.path.abspath(d)) |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
229 try: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
230 #Check path access |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
231 os.mkdir(d) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
232 except Exception, err: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
233 log_manifest(manifest) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
234 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
|
235 sys.exit(1) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
236 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
237 #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
|
238 #print cmd |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
239 |
20
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
240 if out_log and out_log != "-": |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
241 handle = open(out_log, "w") |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
242 else: |
aeb3e35f8236
Uploaded v0.0.4 preview, made MAF and BAM output optional
peterjc
parents:
19
diff
changeset
|
243 handle = open(os.devnull, "w") |
4 | 244 handle.write("======================== MIRA manifest (instructions) ========================\n") |
245 m = open(manifest, "rU") | |
246 for line in m: | |
247 handle.write(line) | |
248 m.close() | |
249 del m | |
250 handle.write("\n") | |
251 handle.write("============================ Starting MIRA now ===============================\n") | |
252 handle.flush() | |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
253 try: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
254 #Run MIRA |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
255 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
|
256 stdout=handle, |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
257 stderr=subprocess.STDOUT) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
258 except Exception, err: |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
259 log_manifest(manifest) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
260 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
|
261 #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
|
262 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
|
263 handle.close() |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
264 sys.exit(1) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
265 #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
|
266 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
|
267 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
|
268 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
|
269 return_code = child.returncode |
4 | 270 handle.write("\n") |
271 handle.write("============================ MIRA has finished ===============================\n") | |
272 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
|
273 if return_code: |
9 | 274 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
|
275 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
|
276 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
|
277 handle.close() |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
278 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
|
279 log_manifest(manifest) |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
280 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
|
281 return_code) |
4 | 282 handle.flush() |
283 | |
284 if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"): | |
285 handle.write("\n") | |
286 handle.write("====================== Extract Large Contigs failed ==========================\n") | |
287 e = open("MIRA_assembly/MIRA_d_results/ec.log", "rU") | |
288 for line in e: | |
289 handle.write(line) | |
290 e.close() | |
291 handle.write("============================ (end of ec.log) =================================\n") | |
292 handle.flush() | |
0
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
293 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
294 #print "Collecting output..." |
9 | 295 start_time = time.time() |
4 | 296 collect_output(temp, name, handle) |
9 | 297 collect_time = time.time() - start_time |
298 handle.write("MIRA took %0.2f hours; collecting output %0.2f minutes\n" % (run_time / 3600.0, collect_time / 60.0)) | |
299 print("MIRA took %0.2f hours; collecting output %0.2f minutes\n" % (run_time / 3600.0, collect_time / 60.0)) | |
4 | 300 |
301 if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"): | |
302 #Treat as an error, but doing this AFTER collect_output | |
303 sys.stderr.write("Extract Large Contigs failed\n") | |
304 handle.write("Extract Large Contigs failed\n") | |
305 handle.close() | |
306 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
|
307 |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
308 #print "Cleaning up..." |
32f693f6e741
Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff
changeset
|
309 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
|
310 |
9 | 311 handle.write("\nDone\n") |
4 | 312 handle.close() |
313 print("Done") |