annotate tools/mira4_0/mira4_bait.py @ 4:1713289d9908 draft default tip

v0.0.11 tweak for use with bioconda dependencies
author peterjc
date Thu, 10 Aug 2017 11:09:10 -0400
parents 4eb32a3d67d1
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
2 """A simple wrapper script to call MIRA4's mirabait and collect its output.
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
3 """
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
5 from __future__ import print_function
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
6
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
7 import os
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
8 import shutil
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
9 import subprocess
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
10 import sys
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
11 import time
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
12
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
13 WRAPPER_VER = "0.0.11" # Keep in sync with the XML file
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
14
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
15
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
16 def get_version(mira_binary):
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
17 """Run MIRA to find its version number"""
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
18 # At the commend line I would use: mira -v | head -n 1
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
19 # however there is some pipe error when doing that here.
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
20 cmd = [mira_binary, "-v"]
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
21 try:
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
22 child = subprocess.Popen(cmd, universal_newlines=True,
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
23 stdout=subprocess.PIPE,
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
24 stderr=subprocess.STDOUT)
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
25 except Exception as err:
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
26 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
27 sys.exit(1)
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
28 ver, tmp = child.communicate()
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
29 del child
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
30 # Workaround for -v not working in mirabait 4.0RC4
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
31 if "invalid option" in ver.split("\n", 1)[0]:
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
32 for line in ver.split("\n", 1):
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
33 if " version " in line:
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
34 line = line.split()
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
35 return line[line.index("version") + 1].rstrip(")")
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
36 sys.exit("Could not determine MIRA version:\n%s" % ver)
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
37 return ver.split("\n", 1)[0]
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
38
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
39
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
40 if "MIRA4" in os.environ:
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
41 mira_path = os.environ["MIRA4"]
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
42 mira_binary = os.path.join(mira_path, "mirabait")
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
43 if not os.path.isfile(mira_binary):
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
44 sys.exit("Missing mirabait under $MIRA4, %r\nFolder contained: %s"
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
45 % (mira_binary, ", ".join(os.listdir(mira_path))))
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
46 else:
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
47 sys.stderr.write("DEBUG: Since $MIRA4 is not set, assuming mira binaries are on $PATH.\n")
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
48 mira_path = None
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
49 mira_binary = "mirabait"
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
50
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
51 mira_ver = get_version(mira_binary)
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
52 if not mira_ver.strip().startswith("4.0"):
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
53 sys.exit("This wrapper is for MIRA V4.0, not:\n%s" % mira_ver)
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
54 if "-v" in sys.argv or "--version" in sys.argv:
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
55 print("%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER))
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
56 sys.exit(0)
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
57
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
58
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
59 format, output_choice, strand_choice, kmer_length, min_occurance, bait_file, in_file, out_file = sys.argv[1:]
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
60
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
61 if format.startswith("fastq"):
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
62 format = "fastq"
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
63 elif format == "mira":
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
64 format = "maf"
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
65 elif format != "fasta":
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
66 sys.exit("Was not expected format %r" % format)
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
67
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
68 assert out_file.endswith(".dat")
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
69 out_file_stem = out_file[:-4]
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
70
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
71 cmd_list = [mira_binary, "-f", format, "-t", format,
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
72 "-k", kmer_length, "-n", min_occurance,
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
73 bait_file, in_file, out_file_stem]
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
74 if output_choice == "pos":
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
75 pass
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
76 elif output_choice == "neg":
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
77 # Invert the selection...
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
78 cmd_list.insert(1, "-i")
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
79 else:
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
80 sys.exit("Output choice should be 'pos' or 'neg', not %r" % output_choice)
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
81 if strand_choice == "both":
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
82 pass
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
83 elif strand_choice == "fwd":
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
84 # Ingore reverse strand...
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
85 cmd_list.insert(1, "-r")
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
86 else:
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
87 sys.exit("Strand choice should be 'both' or 'fwd', not %r" % strand_choice)
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
88
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
89 cmd = " ".join(cmd_list)
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
90 # print cmd
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
91 start_time = time.time()
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
92 try:
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
93 # Run MIRA
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
94 child = subprocess.Popen(cmd_list, universal_newlines=True,
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
95 stdout=subprocess.PIPE,
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
96 stderr=subprocess.STDOUT)
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
97 except Exception as err:
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
98 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
99 sys.exit(1)
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
100 # Use .communicate as can get deadlocks with .wait(),
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
101 stdout, stderr = child.communicate()
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
102 assert stderr is None # Due to way we ran with subprocess
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
103 run_time = time.time() - start_time
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
104 return_code = child.returncode
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
105 print("mirabait took %0.2f minutes" % (run_time / 60.0))
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
106
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
107 if return_code:
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
108 sys.stderr.write(stdout)
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
109 sys.exit("Return error code %i from command:\n%s" % (return_code, cmd),
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
110 return_code)
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
111
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
112 # Capture output
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
113 out_tmp = out_file_stem + "." + format
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
114 if not os.path.isfile(out_tmp):
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
115 sys.stderr.write(stdout)
4
1713289d9908 v0.0.11 tweak for use with bioconda dependencies
peterjc
parents: 2
diff changeset
116 sys.exit("Missing output file from mirabait: %s" % out_tmp)
2
4eb32a3d67d1 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
117 shutil.move(out_tmp, out_file)