annotate tools/mira4/mira4.py @ 10:79759fdec6cb draft

Uploaded v0.0.2 preview 2, attempt auto-install for 64 bit Mac OS X.
author peterjc
date Tue, 10 Dec 2013 06:26:19 -0500
parents 302d13490b23
children 02350bef2e99
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
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
9
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
10 #Do we need any PYTHONPATH magic?
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
11 from mira4_make_bam import make_bam
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
12
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
13 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
14
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
15 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
16 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
17 sys.exit(err)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
18
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 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
21 """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
22 # 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
23 # 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
24 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
25 try:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
26 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
27 stdout=subprocess.PIPE,
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
28 stderr=subprocess.STDOUT)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
29 except Exception, err:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
30 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
31 sys.exit(1)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
32 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
33 del child
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
34 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
35
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
36
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
37 try:
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
38 mira_path = os.environ["MIRA4"]
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
39 except ImportError:
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
40 stop_err("Environment variable $MIRA4 not set")
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
41 mira_binary = os.path.join(mira_path, "mira")
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
42 if not os.path.isfile(mira_binary):
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
43 stop_err("Missing mira under $MIRA4, %r" % mira_binary)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
44 mira_convert = os.path.join(mira_path, "miraconvert")
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
45 if not os.path.isfile(mira_convert):
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
46 stop_err("Missing miraconvert under $MIRA4, %r" % mira_convert)
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
47
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
48 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
49 if not mira_ver.strip().startswith("4.0"):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
50 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
51 mira_convert_ver = get_version(mira_convert)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
52 if not mira_convert_ver.strip().startswith("4.0"):
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_convert))
5
ffefb87bd414 Uploaded v0.0.1 preview 5, using MIRA 4.0 RC4, supports segment_placement (pairing type)
peterjc
parents: 4
diff changeset
54 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
55 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
56 if mira_ver != mira_convert_ver:
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
57 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
58 sys.exit(0)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
59
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
60 def fix_threads(manifest):
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
61 """Tweak the manifest to alter the number of threads."""
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
62 try:
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
63 threads = int(os.environ.get("GALAXY_SLOTS", "1"))
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
64 except ValueError:
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
65 threads = 1
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
66 assert 1 <= threads
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
67 if threads == 1:
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
68 #Nothing to do...
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
69 return
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
70
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
71 handle = open(manifest)
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
72 text = handle.read()
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
73 handle.close()
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
74
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
75 text = text.replace(" -GE:not=1 ", " -GE:not=%i " % threads)
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
76
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
77 handle = open(manifest, "w")
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
78 handle.write(text)
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
79 handle.flush()
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
80 handle.close()
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
81
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
82 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
83 """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
84 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
85 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
86 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
87 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
88 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
89
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
90
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
91 def collect_output(temp, name, handle):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
92 """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
93 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
94 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
95 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
96 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
97 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
98 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
99 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
100 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
101 missing = []
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
102
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
103 old_maf = "%s/%s_out.maf" % (f, name)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
104 if not os.path.isfile(old_maf):
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
105 #Triggered extractLargeContigs.sh?
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
106 old_maf = "%s/%s_LargeContigs_out.maf" % (f, name)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
107
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
108 #De novo or single strain mapping,
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
109 old_fasta = "%s/%s_out.unpadded.fasta" % (f, name)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
110 ref_fasta = "%s/%s_out.padded.fasta" % (f, name)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
111 if not os.path.isfile(old_fasta):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
112 #Mapping (StrainX versus reference) or de novo
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
113 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
114 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
115 if not os.path.isfile(old_fasta):
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
116 old_fasta = "%s/%s_out_ReferenceStrain.unpadded.fasta" % (f, name)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
117 ref_fasta = "%s/%s_out_ReferenceStrain.padded.fasta" % (f, name)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
118
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
119
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
120 missing = False
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
121 for old, new in [(old_maf, out_maf),
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
122 (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
123 if not os.path.isfile(old):
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
124 missing = True
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
125 else:
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
126 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
127 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
128 if missing:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
129 log_manifest(manifest)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
130 sys.stderr.write("Contents of %r:\n" % f)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
131 for filename in sorted(os.listdir(f)):
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
132 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
133
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
134 #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
135 #using the reference FASTA file...
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
136 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
137 if msg:
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
138 stop_err(msg)
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
139
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
140 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
141 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
142 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
143 shutil.rmtree(folder)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
144
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
145 #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
146 #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
147 #/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
148 temp = "."
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
149 #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
150 name = "MIRA"
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
151 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
152
7
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
153 fix_threads(manifest)
902f01c1084b Uploaded v0.0.1 preview 7, with mirabait wrapper
peterjc
parents: 5
diff changeset
154
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
155 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
156 #cmd_list =sys.argv[8:]
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
157 cmd_list = [mira_binary, manifest]
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
158 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
159
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
160 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
161 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
162 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
163 try:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
164 #Check path access
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
165 os.mkdir(d)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
166 except Exception, err:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
167 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
168 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
169 sys.exit(1)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
170
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
171 #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
172 #print cmd
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
173
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
174 handle = open(out_log, "w")
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
175 handle.write("======================== MIRA manifest (instructions) ========================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
176 m = open(manifest, "rU")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
177 for line in m:
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
178 handle.write(line)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
179 m.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
180 del m
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
181 handle.write("\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
182 handle.write("============================ Starting MIRA now ===============================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
183 handle.flush()
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
184 try:
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
185 #Run MIRA
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
186 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
187 stdout=handle,
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
188 stderr=subprocess.STDOUT)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
189 except Exception, err:
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)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
191 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
192 #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
193 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
194 handle.close()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
195 sys.exit(1)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
196 #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
197 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
198 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
199 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
200 return_code = child.returncode
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
201 handle.write("\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
202 handle.write("============================ MIRA has finished ===============================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
203 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
204 if return_code:
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
205 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
206 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
207 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
208 handle.close()
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
209 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
210 log_manifest(manifest)
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
211 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
212 return_code)
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
213 handle.flush()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
214
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
215 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
216 handle.write("\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
217 handle.write("====================== Extract Large Contigs failed ==========================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
218 e = open("MIRA_assembly/MIRA_d_results/ec.log", "rU")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
219 for line in e:
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
220 handle.write(line)
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
221 e.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
222 handle.write("============================ (end of ec.log) =================================\n")
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
223 handle.flush()
0
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
224
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
225 #print "Collecting output..."
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
226 start_time = time.time()
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
227 collect_output(temp, name, handle)
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
228 collect_time = time.time() - start_time
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
229 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
230 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
231
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
232 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
233 #Treat as an error, but doing this AFTER collect_output
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
234 sys.stderr.write("Extract Large Contigs failed\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 handle.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
237 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
238
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
239 #print "Cleaning up..."
32f693f6e741 Uploaded v0.0.1 preview0, very much a work in progress, primarily checking mira_datatypes dependency
peterjc
parents:
diff changeset
240 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
241
9
302d13490b23 Uploaded v0.0.2 preview 1, BAM output
peterjc
parents: 7
diff changeset
242 handle.write("\nDone\n")
4
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
243 handle.close()
df86ed992a1b Uploaded preview 4, lots of work on mapping
peterjc
parents: 0
diff changeset
244 print("Done")