annotate tools/mira4_0/mira4_bait.py @ 38:cee8f9005e43 draft

planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit 206259620376b322fc8ed99a6efdd3712f38764b
author peterjc
date Wed, 11 Jul 2018 12:35:35 -0400
parents 259891fce7fd
children bbf14bb9607b
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
38
cee8f9005e43 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit 206259620376b322fc8ed99a6efdd3712f38764b
peterjc
parents: 35
diff changeset
2 """A simple wrapper script to call MIRA4's mirabait and collect its output."""
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
3
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
4 from __future__ import print_function
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
5
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
6 import os
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
7 import shutil
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
8 import subprocess
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
9 import sys
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
10 import time
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
11
35
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
12 WRAPPER_VER = "0.0.11" # Keep in sync with the XML file
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
13
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
14
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
15 def get_version(mira_binary):
38
cee8f9005e43 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit 206259620376b322fc8ed99a6efdd3712f38764b
peterjc
parents: 35
diff changeset
16 """Run MIRA to find its version number."""
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
17 # 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
18 # however there is some pipe error when doing that here.
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
19 cmd = [mira_binary, "-v"]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
20 try:
34
0785a6537f3e planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit 6405ba93fcec7ea93452bf54d559c7507ee7a57c
peterjc
parents: 32
diff changeset
21 child = subprocess.Popen(cmd, universal_newlines=True,
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
22 stdout=subprocess.PIPE,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
23 stderr=subprocess.STDOUT)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
24 except Exception as err:
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
25 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
26 sys.exit(1)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
27 ver, tmp = child.communicate()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
28 del child
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
29 # Workaround for -v not working in mirabait 4.0RC4
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
30 if "invalid option" in ver.split("\n", 1)[0]:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
31 for line in ver.split("\n", 1):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
32 if " version " in line:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
33 line = line.split()
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
34 return line[line.index("version") + 1].rstrip(")")
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
35 sys.exit("Could not determine MIRA version:\n%s" % ver)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
36 return ver.split("\n", 1)[0]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
37
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
38
35
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
39 if "MIRA4" in os.environ:
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
40 mira_path = os.environ["MIRA4"]
35
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
41 mira_binary = os.path.join(mira_path, "mirabait")
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
42 if not os.path.isfile(mira_binary):
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
43 sys.exit("Missing mirabait under $MIRA4, %r\nFolder contained: %s"
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
44 % (mira_binary, ", ".join(os.listdir(mira_path))))
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
45 else:
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
46 sys.stderr.write("DEBUG: Since $MIRA4 is not set, assuming mira binaries are on $PATH.\n")
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
47 mira_path = None
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
48 mira_binary = "mirabait"
259891fce7fd planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit e4c56df75150c82d3e9c4ac487c4209108e52412
peterjc
parents: 34
diff changeset
49
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
50 mira_ver = get_version(mira_binary)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
51 if not mira_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
52 sys.exit("This wrapper is for MIRA V4.0, not:\n%s" % mira_ver)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
53 if "-v" in sys.argv or "--version" in sys.argv:
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
54 print("%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
55 sys.exit(0)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
56
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
57
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
58 format, output_choice, strand_choice, kmer_length, min_occurance, bait_file, in_file, out_file = sys.argv[1:]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
59
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
60 if format.startswith("fastq"):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
61 format = "fastq"
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
62 elif format == "mira":
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
63 format = "maf"
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
64 elif format != "fasta":
31
fd95aaef8818 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit bc3d484c5cd68ddcf456db2fff489d584aa2034c
peterjc
parents: 25
diff changeset
65 sys.exit("Was not expected format %r" % format)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
66
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
67 assert out_file.endswith(".dat")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
68 out_file_stem = out_file[:-4]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
69
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
70 cmd_list = [mira_binary, "-f", format, "-t", format,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
71 "-k", kmer_length, "-n", min_occurance,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
72 bait_file, in_file, out_file_stem]
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
73 if output_choice == "pos":
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
74 pass
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
75 elif output_choice == "neg":
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
76 # Invert the selection...
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
77 cmd_list.insert(1, "-i")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
78 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
79 sys.exit("Output choice should be 'pos' or 'neg', not %r" % output_choice)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
80 if strand_choice == "both":
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
81 pass
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
82 elif strand_choice == "fwd":
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
83 # Ingore reverse strand...
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
84 cmd_list.insert(1, "-r")
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
85 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
86 sys.exit("Strand choice should be 'both' or 'fwd', not %r" % strand_choice)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
87
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
88 cmd = " ".join(cmd_list)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
89 # print cmd
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
90 start_time = time.time()
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
91 try:
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
92 # Run MIRA
34
0785a6537f3e planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit 6405ba93fcec7ea93452bf54d559c7507ee7a57c
peterjc
parents: 32
diff changeset
93 child = subprocess.Popen(cmd_list, universal_newlines=True,
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
94 stdout=subprocess.PIPE,
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
95 stderr=subprocess.STDOUT)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
96 except Exception as err:
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
97 sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
98 sys.exit(1)
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
99 # Use .communicate as can get deadlocks with .wait(),
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
100 stdout, stderr = child.communicate()
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
101 assert stderr is None # Due to way we ran with subprocess
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
102 run_time = time.time() - start_time
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
103 return_code = child.returncode
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
104 print("mirabait took %0.2f minutes" % (run_time / 60.0))
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
105
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
106 if return_code:
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
107 sys.stderr.write(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
108 sys.exit("Return error code %i from command:\n%s" % (return_code, cmd),
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
109 return_code)
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
110
32
56b421d59805 planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6
peterjc
parents: 31
diff changeset
111 # Capture output
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
112 out_tmp = out_file_stem + "." + format
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
113 if not os.path.isfile(out_tmp):
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
114 sys.stderr.write(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
115 sys.exit("Missing output file from mirabait: %s" % out_tmp)
25
55ae131c5862 v0.0.8 - renamed folder, added note about mirabait
peterjc
parents:
diff changeset
116 shutil.move(out_tmp, out_file)