Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/tmhmm2.py @ 2:747cec3192d3

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
author peterjc
date Tue, 07 Jun 2011 17:38:43 -0400
parents 9a8a7f680dd6
children 5e62aefb2918
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
2 """Wrapper for TMHMM v2.0 for use in Galaxy.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
3
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
4 This script takes exactly two command line arguments - an input protein FASTA
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
5 filename and an output tabular filename. It then calls the standalone TMHMM
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
6 v2.0 program (not the webservice) requesting the short output (one line per
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
7 protein).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
8
2
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
9 The first major feature is cleaning up the tabular output. The short form raw
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
10 output from TMHMM v2.0 looks like this (six columns tab separated):
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
11
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
12 gi|2781234|pdb|1JLY|B len=304 ExpAA=0.01 First60=0.00 PredHel=0 Topology=o
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
13 gi|4959044|gb|AAD34209.1|AF069992_1 len=600 ExpAA=0.00 First60=0.00 PredHel=0 Topology=o
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
14 gi|671626|emb|CAA85685.1| len=473 ExpAA=0.19 First60=0.00 PredHel=0 Topology=o
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
15 gi|3298468|dbj|BAA31520.1| len=107 ExpAA=59.37 First60=31.17 PredHel=3 Topology=o23-45i52-74o89-106i
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
16
2
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
17 If there are any additional 'comment' lines starting with the hash (#)
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
18 character these are ignored by this script.
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
19
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
20 In order to make it easier to use in Galaxy, this wrapper script simplifies
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
21 this to remove the redundant tags, and instead adds a comment line at the
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
22 top with the column names:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
23
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
24 #ID len ExpAA First60 PredHel Topology
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
25 gi|2781234|pdb|1JLY|B 304 0.01 60 0.00 0 o
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
26 gi|4959044|gb|AAD34209.1|AF069992_1 600 0.00 0 0.00 0 o
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
27 gi|671626|emb|CAA85685.1| 473 0.19 0.00 0 o
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
28 gi|3298468|dbj|BAA31520.1| 107 59.37 31.17 3 o23-45i52-74o89-106i
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
29
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
30 The second major potential feature is taking advantage of multiple cores
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
31 (since TMHMM v2.0 itself is single threaded) by dividing the input FASTA file
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
32 into chunks and running multiple copies of TMHMM in parallel. I would normally
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
33 use Python's multiprocessing library in this situation but it requires at
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
34 least Python 2.6 and at the time of writing Galaxy still supports Python 2.4.
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
35
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
36 Also tmhmm2 can fail without returning an error code, for example if run on a
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
37 64 bit machine with only the 32 bit binaries installed. This script will spot
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
38 when there is no output from tmhmm2, and raise an error.
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
39 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
40 import sys
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
41 import os
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
42 from seq_analysis_utils import stop_err, split_fasta, run_jobs
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
43
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
44 FASTA_CHUNK = 500
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
45
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
46 if len(sys.argv) != 4:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
47 stop_err("Require three arguments, number of threads (int), input protein FASTA file & output tabular file")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
48 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
49 num_threads = int(sys.argv[1])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
50 except:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
51 num_threads = 0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
52 if num_threads < 1:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
53 stop_err("Threads argument %s is not a positive integer" % sys.argv[1])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
54 fasta_file = sys.argv[2]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
55 tabular_file = sys.argv[3]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
56
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
57 def clean_tabular(raw_handle, out_handle):
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
58 """Clean up tabular TMHMM output, returns output line count."""
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
59 count = 0
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
60 for line in raw_handle:
2
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
61 if not line.strip() or line.startswith("#"):
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
62 #Ignore any blank lines or comment lines
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
63 continue
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
64 parts = line.rstrip("\r\n").split("\t")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
65 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
66 identifier, length, expAA, first60, predhel, topology = parts
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
67 except:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
68 assert len(parts)!=6
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
69 stop_err("Bad line: %r" % line)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
70 assert length.startswith("len="), line
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
71 length = length[4:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
72 assert expAA.startswith("ExpAA="), line
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
73 expAA = expAA[6:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
74 assert first60.startswith("First60="), line
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 first60 = first60[8:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
76 assert predhel.startswith("PredHel="), line
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
77 predhel = predhel[8:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 assert topology.startswith("Topology="), line
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 topology = topology[9:]
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
80 out_handle.write("%s\t%s\t%s\t%s\t%s\t%s\n" \
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
81 % (identifier, length, expAA, first60, predhel, topology))
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
82 count += 1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
83 return count
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
84
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
85 #Note that if the input FASTA file contains no sequences,
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
86 #split_fasta returns an empty list (i.e. zero temp files).
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
87 fasta_files = split_fasta(fasta_file, tabular_file, FASTA_CHUNK)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
88 temp_files = [f+".out" for f in fasta_files]
2
747cec3192d3 Migrated tool version 0.0.5 from old tool shed archive to new tool shed repository
peterjc
parents: 1
diff changeset
89 jobs = ["tmhmm -short %s > %s" % (fasta, temp)
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
90 for fasta, temp in zip(fasta_files, temp_files)]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
91
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
92 def clean_up(file_list):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
93 for f in file_list:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
94 if os.path.isfile(f):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
95 os.remove(f)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
96
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
97 if len(jobs) > 1 and num_threads > 1:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
98 #A small "info" message for Galaxy to show the user.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
99 print "Using %i threads for %i tasks" % (min(num_threads, len(jobs)), len(jobs))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
100 results = run_jobs(jobs, num_threads)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
101 for fasta, temp, cmd in zip(fasta_files, temp_files, jobs):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
102 error_level = results[cmd]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
103 if error_level:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
104 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
105 output = open(temp).readline()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
106 except IOError:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
107 output = ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
108 clean_up(fasta_files)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
109 clean_up(temp_files)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
110 stop_err("One or more tasks failed, e.g. %i from %r gave:\n%s" % (error_level, cmd, output),
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
111 error_level)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
112 del results
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
113 del jobs
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
114
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
115 out_handle = open(tabular_file, "w")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
116 out_handle.write("#ID\tlen\tExpAA\tFirst60\tPredHel\tTopology\n")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
117 for temp in temp_files:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
118 data_handle = open(temp)
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
119 count = clean_tabular(data_handle, out_handle)
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
120 data_handle.close()
1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
121 if not count:
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
122 clean_up(fasta_files)
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
123 clean_up(temp_files)
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
124 stop_err("No output from tmhmm2")
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
125 out_handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
126
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
127 clean_up(fasta_files)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
128 clean_up(temp_files)