annotate tools/mira4_0/mira4_convert.py @ 31:fd95aaef8818 draft

planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
author peterjc
date Wed, 10 Feb 2016 09:07:39 -0500
parents 55ae131c5862
children 56b421d59805
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
2 """A simple wrapper script to call MIRA and collect its output.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
3
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
4 This focuses on the miraconvert binary.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
5 """
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
6 import os
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
7 import sys
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
8 import subprocess
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
9 import shutil
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
10 from optparse import OptionParser
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
11 try:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
12 from io import BytesIO
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
13 except ImportError:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
14 #Should we worry about Python 2.5 or older?
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
15 from StringIO import StringIO as BytesIO
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
16
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
17 #Do we need any PYTHONPATH magic?
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
18 from mira4_make_bam import depad
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
19
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
20 WRAPPER_VER = "0.0.7" # Keep in sync with the XML file
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
21
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
22
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
23 def run(cmd):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
24 #Avoid using shell=True when we call subprocess to ensure if the Python
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
25 #script is killed, so too is the child process.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
26 try:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
27 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
28 except Exception, err:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
29 sys.exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
30 #Use .communicate as can get deadlocks with .wait(),
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
31 stdout, stderr = child.communicate()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
32 return_code = child.returncode
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
33 if return_code:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
34 cmd_str = " ".join(cmd) # doesn't quote spaces etc
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
35 if stderr and stdout:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
36 sys.exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
37 else:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
38 sys.exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
39
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
40 def get_version(mira_binary):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
41 """Run MIRA to find its version number"""
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
42 # At the commend line I would use: mira -v | head -n 1
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
43 # however there is some pipe error when doing that here.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
44 cmd = [mira_binary, "-v"]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
45 try:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
46 child = subprocess.Popen(cmd,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
47 stdout=subprocess.PIPE,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
48 stderr=subprocess.STDOUT)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
49 except Exception, err:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
50 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
51 sys.exit(1)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
52 ver, tmp = child.communicate()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
53 del child
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
54 return ver.split("\n", 1)[0].strip()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
55
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
56 #Parse Command Line
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
57 usage = """Galaxy MIRA4 wrapper script v%s - use as follows:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
58
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
59 $ python mira4_convert.py ...
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
60
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
61 This will run the MIRA miraconvert binary and collect its output files as directed.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
62 """ % WRAPPER_VER
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
63 parser = OptionParser(usage=usage)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
64 parser.add_option("--input", dest="input",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
65 default=None, metavar="FILE",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
66 help="MIRA input filename")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
67 parser.add_option("-x", "--min_length", dest="min_length",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
68 default="0",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
69 help="Minimum contig length")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
70 parser.add_option("-y", "--min_cover", dest="min_cover",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
71 default="0",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
72 help="Minimum average contig coverage")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
73 parser.add_option("-z", "--min_reads", dest="min_reads",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
74 default="0",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
75 help="Minimum reads per contig")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
76 parser.add_option("--maf", dest="maf",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
77 default="", metavar="FILE",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
78 help="MIRA MAF output filename")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
79 parser.add_option("--ace", dest="ace",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
80 default="", metavar="FILE",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
81 help="ACE output filename")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
82 parser.add_option("--bam", dest="bam",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
83 default="", metavar="FILE",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
84 help="Unpadded BAM output filename")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
85 parser.add_option("--fasta", dest="fasta",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
86 default="", metavar="FILE",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
87 help="Unpadded FASTA output filename")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
88 parser.add_option("--cstats", dest="cstats",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
89 default="", metavar="FILE",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
90 help="Contig statistics filename")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
91 parser.add_option("-v", "--version", dest="version",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
92 default=False, action="store_true",
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
93 help="Show version and quit")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
94 options, args = parser.parse_args()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
95 if args:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
96 sys.exit("Expected options (e.g. --input example.maf), not arguments")
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
97
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
98 input_maf = options.input
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
99 out_maf = options.maf
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
100 out_bam = options.bam
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
101 out_fasta = options.fasta
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
102 out_ace = options.ace
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
103 out_cstats = options.cstats
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
104
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
105 try:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
106 mira_path = os.environ["MIRA4"]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
107 except KeyError:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
108 sys.exit("Environment variable $MIRA4 not set")
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
109 mira_convert = os.path.join(mira_path, "miraconvert")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
110 if not os.path.isfile(mira_convert):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
111 sys.exit("Missing miraconvert under $MIRA4, %r\nFolder contained: %s"
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
112 % (mira_convert, ", ".join(os.listdir(mira_path))))
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
113
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
114 mira_convert_ver = get_version(mira_convert)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
115 if not mira_convert_ver.strip().startswith("4.0"):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
116 sys.exit("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_convert_ver, mira_convert))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
117 if options.version:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
118 print("%s, MIRA wrapper version %s" % (mira_convert_ver, WRAPPER_VER))
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
119 sys.exit(0)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
120
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
121 if not input_maf:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
122 sys.exit("Input MIRA file is required")
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
123 elif not os.path.isfile(input_maf):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
124 sys.exit("Missing input MIRA file: %r" % input_maf)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
125
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
126 if not (out_maf or out_bam or out_fasta or out_ace or out_cstats):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
127 sys.exit("No output requested")
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
128
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
129
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
130 def check_min_int(value, name):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
131 try:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
132 i = int(value)
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
133 except ValueError:
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
134 sys.exit("Bad %s setting, %r" % (name, value))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
135 if i < 0:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
136 sys.exit("Negative %s setting, %r" % (name, value))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
137 return i
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
138
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
139 min_length = check_min_int(options.min_length, "minimum length")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
140 min_cover = check_min_int(options.min_cover, "minimum cover")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
141 min_reads = check_min_int(options.min_reads, "minimum reads")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
142
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
143 #TODO - Run MIRA in /tmp or a configurable directory?
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
144 #Currently Galaxy puts us somewhere safe like:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
145 #/opt/galaxy-dist/database/job_working_directory/846/
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
146 temp = "."
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
147
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
148
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
149 cmd_list = [mira_convert]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
150 if min_length:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
151 cmd_list.extend(["-x", str(min_length)])
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
152 if min_cover:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
153 cmd_list.extend(["-y", str(min_cover)])
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
154 if min_reads:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
155 cmd_list.extend(["-z", str(min_reads)])
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
156 cmd_list.extend(["-f", "maf", input_maf, os.path.join(temp, "converted")])
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
157 if out_maf:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
158 cmd_list.append("maf")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
159 if out_bam:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
160 cmd_list.append("samnbb")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
161 if not out_fasta:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
162 #Need this for samtools depad
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
163 out_fasta = os.path.join(temp, "depadded.fasta")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
164 if out_fasta:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
165 cmd_list.append("fasta")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
166 if out_ace:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
167 cmd_list.append("ace")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
168 if out_cstats:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
169 cmd_list.append("cstats")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
170 run(cmd_list)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
171
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
172 def collect(old, new):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
173 if not os.path.isfile(old):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
174 sys.exit("Missing expected output file %s" % old)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
175 shutil.move(old, new)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
176
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
177 if out_maf:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
178 collect(os.path.join(temp, "converted.maf"), out_maf)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
179 if out_fasta:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
180 #Can we look at the MAF file to see if there are multiple strains?
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
181 old = os.path.join(temp, "converted_AllStrains.unpadded.fasta")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
182 if os.path.isfile(old):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
183 collect(old, out_fasta)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
184 else:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
185 #Might the output be filtered down to zero contigs?
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
186 old = os.path.join(temp, "converted.fasta")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
187 if not os.path.isfile(old):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
188 sys.exit("Missing expected output FASTA file")
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
189 elif os.path.getsize(old) == 0:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
190 print("Warning - no contigs (harsh filters?)")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
191 collect(old, out_fasta)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
192 else:
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
193 sys.exit("Missing expected output FASTA file (only generic file present)")
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
194 if out_ace:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
195 collect(os.path.join(temp, "converted.maf"), out_ace)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
196 if out_cstats:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
197 collect(os.path.join(temp, "converted_info_contigstats.txt"), out_cstats)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
198
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
199 if out_bam:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
200 assert os.path.isfile(out_fasta)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
201 old = os.path.join(temp, "converted.samnbb")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
202 if not os.path.isfile(old):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
203 old = os.path.join(temp, "converted.sam")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
204 if not os.path.isfile(old):
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
205 sys.exit("Missing expected intermediate file %s" % old)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
206 h = BytesIO()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
207 msg = depad(out_fasta, old, out_bam, h)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
208 if msg:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
209 print(msg)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
210 print(h.getvalue())
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
211 h.close()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
212 sys.exit(1)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
213 h.close()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
214 if out_fasta == os.path.join(temp, "depadded.fasta"):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
215 #Not asked for by Galaxy, no longer needed
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
216 os.remove(out_fasta)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
217
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
218 if min_length or min_cover or min_reads:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
219 print("Filtered.")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
220 else:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
221 print("Converted.")