comparison tools/mira4/mira4_bait.py @ 0:6a88b42ce6b9 draft

Uploaded v0.0.4, previously only on the TestToolShed
author peterjc
date Fri, 21 Nov 2014 06:42:56 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:6a88b42ce6b9
1 #!/usr/bin/env python
2 """A simple wrapper script to call MIRA4's mirabait and collect its output.
3 """
4 import os
5 import sys
6 import subprocess
7 import shutil
8 import time
9
10 WRAPPER_VER = "0.0.1" #Keep in sync with the XML file
11
12 def stop_err(msg, err=1):
13 sys.stderr.write(msg+"\n")
14 sys.exit(err)
15
16
17 def get_version(mira_binary):
18 """Run MIRA to find its version number"""
19 # At the commend line I would use: mira -v | head -n 1
20 # however there is some pipe error when doing that here.
21 cmd = [mira_binary, "-v"]
22 try:
23 child = subprocess.Popen(cmd,
24 stdout=subprocess.PIPE,
25 stderr=subprocess.STDOUT)
26 except Exception, err:
27 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
28 sys.exit(1)
29 ver, tmp = child.communicate()
30 del child
31 #Workaround for -v not working in mirabait 4.0RC4
32 if "invalid option" in ver.split("\n", 1)[0]:
33 for line in ver.split("\n", 1):
34 if " version " in line:
35 line = line.split()
36 return line[line.index("version")+1].rstrip(")")
37 stop_err("Could not determine MIRA version:\n%s" % ver)
38 return ver.split("\n", 1)[0]
39
40 try:
41 mira_path = os.environ["MIRA4"]
42 except KeyError:
43 stop_err("Environment variable $MIRA4 not set")
44 mira_binary = os.path.join(mira_path, "mirabait")
45 if not os.path.isfile(mira_binary):
46 stop_err("Missing mirabait under $MIRA4, %r\nFolder contained: %s"
47 % (mira_binary, ", ".join(os.listdir(mira_path))))
48 mira_ver = get_version(mira_binary)
49 if not mira_ver.strip().startswith("4.0"):
50 stop_err("This wrapper is for MIRA V4.0, not:\n%s" % mira_ver)
51 if "-v" in sys.argv or "--version" in sys.argv:
52 print "%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER)
53 sys.exit(0)
54
55
56 format, output_choice, strand_choice, kmer_length, min_occurance, bait_file, in_file, out_file = sys.argv[1:]
57
58 if format.startswith("fastq"):
59 format = "fastq"
60 elif format == "mira":
61 format = "maf"
62 elif format != "fasta":
63 stop_err("Was not expected format %r" % format)
64
65 assert out_file.endswith(".dat")
66 out_file_stem = out_file[:-4]
67
68 cmd_list = [mira_binary, "-f", format, "-t", format,
69 "-k", kmer_length, "-n", min_occurance,
70 bait_file, in_file, out_file_stem]
71 if output_choice == "pos":
72 pass
73 elif output_choice == "neg":
74 #Invert the selection...
75 cmd_list.insert(1, "-i")
76 else:
77 stop_err("Output choice should be 'pos' or 'neg', not %r" % output_choice)
78 if strand_choice == "both":
79 pass
80 elif strand_choice == "fwd":
81 #Ingore reverse strand...
82 cmd_list.insert(1, "-r")
83 else:
84 stop_err("Strand choice should be 'both' or 'fwd', not %r" % strand_choice)
85
86 cmd = " ".join(cmd_list)
87 #print cmd
88 start_time = time.time()
89 try:
90 #Run MIRA
91 child = subprocess.Popen(cmd_list,
92 stdout=subprocess.PIPE,
93 stderr=subprocess.STDOUT)
94 except Exception, err:
95 log_manifest(manifest)
96 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
97 sys.exit(1)
98 #Use .communicate as can get deadlocks with .wait(),
99 stdout, stderr = child.communicate()
100 assert stderr is None # Due to way we ran with subprocess
101 run_time = time.time() - start_time
102 return_code = child.returncode
103 print "mirabait took %0.2f minutes" % (run_time / 60.0)
104
105 if return_code:
106 sys.stderr.write(stdout)
107 stop_err("Return error code %i from command:\n%s" % (return_code, cmd),
108 return_code)
109
110 #Capture output
111 out_tmp = out_file_stem + "." + format
112 if not os.path.isfile(out_tmp):
113 sys.stderr.write(stdout)
114 stop_err("Missing output file from mirabait: %s" % out_tmp)
115 shutil.move(out_tmp, out_file)