tmhmm_and_signalp: tools/protein_analysis/tmhmm2.py annotate

annotate tools/protein_analysis/tmhmm2.py @ 13:c6e5a737d3a0 draft

Uploaded v0.2.2, was missing some test files

author	peterjc
date	Mon, 22 Apr 2013 11:46:20 -0400
parents	391a142c1e60
children	20139cb4c844

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
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	4 This script takes exactly three command line arguments - number of threads,
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	5 an input protein FASTA filename, and an output tabular filename. It then
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	6 calls the standalone TMHMM v2.0 program (not the webservice) requesting
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	7 the short output (one line per protein).
0 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
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	36 Note that this is somewhat redundant with job-splitting available in Galaxy
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	37 itself (see the SignalP XML file for settings).
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	38
1 9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	39 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	40 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	41 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	42 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	43 import sys
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	44 import os
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	45 import tempfile
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	46 from seq_analysis_utils import stop_err, split_fasta, run_jobs, thread_count
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	47
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	48 FASTA_CHUNK = 500
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	49
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	50 if len(sys.argv) != 4:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	51 stop_err("Require three arguments, number of threads (int), input protein FASTA file & output tabular file")
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	52
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	53 num_threads = thread_count(sys.argv[1], default=4)
0 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
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	57 tmp_dir = tempfile.mkdtemp()
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	58
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	59 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	60 """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	61 count = 0
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	62 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	63 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	64 #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	65 continue
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	66 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	67 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	68 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	69 except:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	70 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	71 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	72 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	73 length = length[4:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	74 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	75 expAA = expAA[6:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	76 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	77 first60 = first60[8:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	78 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	79 predhel = predhel[8:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	80 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	81 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	82 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	83 % (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	84 count += 1
9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	85 return count
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	86
1 9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	87 #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	88 #split_fasta returns an empty list (i.e. zero temp files).
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	89 fasta_files = split_fasta(fasta_file, os.path.join(tmp_dir, "tmhmm"), FASTA_CHUNK)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	90 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	91 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	92 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	93
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	94 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	95 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	96 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	97 os.remove(f)
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	98 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	99 os.rmdir(tmp_dir)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	100 except:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	101 pass
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	102
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	103 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	104 #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	105 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	106 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	107 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	108 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	109 if error_level:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	110 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	111 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	112 except IOError:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	113 output = ""
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	114 clean_up(fasta_files + temp_files)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	115 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	116 error_level)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	117 del results
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	118 del jobs
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	119
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	120 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	121 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	122 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	123 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	124 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	125 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	126 if not count:
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	127 clean_up(fasta_files + temp_files)
1 9a8a7f680dd6 Migrated tool version 0.0.3 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	128 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	129 out_handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	130
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 2 diff changeset	131 clean_up(fasta_files + temp_files)

Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/tmhmm2.py @ 13:c6e5a737d3a0 draft