tmhmm_and_signalp: tools/protein_analysis/signalp3.py annotate

annotate tools/protein_analysis/signalp3.py @ 16:1d5486a02bc4 draft

Uploaded v0.2.4, added unit tests for Promoter 2

author	peterjc
date	Thu, 25 Apr 2013 12:25:23 -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 SignalP v3.0 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
5 ef7ceca37e3f Migrated tool version 0.0.8 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	4 This script takes exactly five command line arguments:
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	5 * the organism type (euk, gram+ or gram-)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	6 * length to truncate sequences to (integer)
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	7 * number of threads to use (integer, defaults to one)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	8 * an input protein FASTA filename
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	9 * output tabular filename.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	10
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	11 There are two further optional arguments
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	12 * cut type (NN_Cmax, NN_Ymax, NN_Smax or HMM_Cmax)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	13 * output GFF3 filename
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	14
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	15 It then calls the standalone SignalP v3.0 program (not the webservice)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	16 requesting the short output (one line per protein) using both NN and HMM
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	17 for predictions.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	18
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	19 First major feature is cleaning up the output. The raw output from SignalP
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	20 v3.0 looks like this (21 columns space separated):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	21
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	22 # SignalP-NN euk predictions # SignalP-HMM euk predictions
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	23 # name Cmax pos ? Ymax pos ? Smax pos ? Smean ? D ? # name ! Cmax pos ? Sprob ?
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	24 gi\|2781234\|pdb\|1JLY\| 0.061 17 N 0.043 17 N 0.199 1 N 0.067 N 0.055 N gi\|2781234\|pdb\|1JLY\|B Q 0.000 17 N 0.000 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	25 gi\|4959044\|gb\|AAD342 0.099 191 N 0.012 38 N 0.023 12 N 0.014 N 0.013 N gi\|4959044\|gb\|AAD34209.1\|AF069992_1 Q 0.000 0 N 0.000 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	26 gi\|671626\|emb\|CAA856 0.139 381 N 0.020 8 N 0.121 4 N 0.067 N 0.044 N gi\|671626\|emb\|CAA85685.1\| Q 0.000 0 N 0.000 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	27 gi\|3298468\|dbj\|BAA31 0.208 24 N 0.184 38 N 0.980 32 Y 0.613 Y 0.398 N gi\|3298468\|dbj\|BAA31520.1\| Q 0.066 24 N 0.139 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	28
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	29 In order to make it easier to use in Galaxy, this wrapper script reformats
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	30 this to use tab separators. Also it removes the redundant truncated name
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	31 column, and assigns unique column names in the header:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	32
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	33 #ID NN_Cmax_score NN_Cmax_pos NN_Cmax_pred NN_Ymax_score NN_Ymax_pos NN_Ymax_pred NN_Smax_score NN_Smax_pos NN_Smax_pred NN_Smean_score NN_Smean_pred NN_D_score NN_D_pred HMM_bang HMM_Cmax_score HMM_Cmax_pos HMM_Cmax_pred HMM_Sprob_score HMM_Sprob_pred
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	34 gi\|2781234\|pdb\|1JLY\|B 0.061 17 N 0.043 17 N 0.199 1 N 0.067 N 0.055 N Q 0.000 17 N 0.000 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	35 gi\|4959044\|gb\|AAD34209.1\|AF069992_1 0.099 191 N 0.012 38 N 0.023 12 N 0.014 N 0.013 N Q 0.000 0 N 0.000 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	36 gi\|671626\|emb\|CAA85685.1\| 0.139 381 N 0.020 8 N 0.121 4 N 0.067 N 0.044 N Q 0.000 0 N 0.000 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	37 gi\|3298468\|dbj\|BAA31520.1\| 0.208 24 N 0.184 38 N 0.980 32 Y 0.613 Y 0.398 N Q 0.066 24 N 0.139 N
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	38
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	39 The second major feature is overcoming SignalP's built in limit of 4000
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	40 sequences by breaking up the input FASTA file into chunks. This also allows
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	41 us to pre-trim the sequences since SignalP only needs their starts.
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 The third major feature is taking advantage of multiple cores (since SignalP
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	44 v3.0 itself is single threaded) by using the individual FASTA input files to
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	45 run multiple copies of TMHMM in parallel. I would normally use Python's
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	46 multiprocessing library in this situation but it requires at least Python 2.6
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	47 and at the time of writing Galaxy still supports Python 2.4.
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	48
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	49 Note that this is somewhat redundant with job-splitting available in Galaxy
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	50 itself (see the SignalP XML file for settings).
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	51
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	52 Finally, you can opt to have a GFF3 file produced which will describe the
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	53 predicted signal peptide and mature peptide for each protein (using one of
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	54 the predictors which gives a cleavage site). WORK IN PROGRESS
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	55 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	56 import sys
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	57 import os
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	58 import tempfile
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	59 from seq_analysis_utils import stop_err, split_fasta, fasta_iterator
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	60 from seq_analysis_utils import 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	61
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	62 FASTA_CHUNK = 500
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	63 MAX_LEN = 6000 #Found by trial and error
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	64
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	65 if len(sys.argv) not in [6,8]:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	66 stop_err("Require five (or 7) arguments, organism, truncate, threads, "
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	67 "input protein FASTA file & output tabular file (plus "
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	68 "optionally cut method and GFF3 output file). "
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	69 "Got %i arguments." % (len(sys.argv)-1))
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	70
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	71 organism = sys.argv[1]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	72 if organism not in ["euk", "gram+", "gram-"]:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	73 stop_err("Organism argument %s is not one of euk, gram+ or gram-" % organism)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	74
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	75 try:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	76 truncate = int(sys.argv[2])
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	77 except:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	78 truncate = 0
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	79 if truncate < 0:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	80 stop_err("Truncate argument %s is not a positive integer (or zero)" % sys.argv[2])
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	81
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	82 num_threads = thread_count(sys.argv[3], default=4)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	83 fasta_file = sys.argv[4]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	84 tabular_file = sys.argv[5]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	85
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	86 if len(sys.argv) == 8:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	87 cut_method = sys.argv[6]
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	88 if cut_method not in ["NN_Cmax", "NN_Ymax", "NN_Smax", "HMM_Cmax"]:
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	89 stop_err("Invalid cut method %r" % cut_method)
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	90 gff3_file = sys.argv[7]
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	91 else:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	92 cut_method = None
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	93 gff3_file = None
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	94
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	95
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	96 tmp_dir = tempfile.mkdtemp()
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	97
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	98 def clean_tabular(raw_handle, out_handle, gff_handle=None, cut_method=None):
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	99 """Clean up SignalP output to make it tabular."""
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	100 if cut_method:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	101 cut_col = {"NN_Cmax" : 2,
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	102 "NN_Ymax" : 5,
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	103 "NN_Smax" : 8,
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	104 "HMM_Cmax" : 16}[cut_method]
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	105 else:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	106 cut_col = None
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	107 for line in raw_handle:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	108 if not line or line.startswith("#"):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	109 continue
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	110 parts = line.rstrip("\r\n").split()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	111 assert len(parts)==21, repr(line)
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	112 assert parts[14].startswith(parts[0]), \
391a142c1e60 Uploaded peterjc parents: 7 diff changeset	113 "Bad entry in SignalP output, ID miss-match:\n%r" % line
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	114 #Remove redundant truncated name column (col 0)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	115 #and put full name at start (col 14)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	116 parts = parts[14:15] + parts[1:14] + parts[15:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	117 out_handle.write("\t".join(parts) + "\n")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	118
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	119 def make_gff(fasta_file, tabular_file, gff_file, cut_method):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	120 cut_col, score_col = {"NN_Cmax" : (2,1),
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	121 "NN_Ymax" : (5,4),
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	122 "NN_Smax" : (8,7),
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	123 "HMM_Cmax" : (16,15),
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	124 }[cut_method]
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	125
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	126 source = "SignalP"
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	127 strand = "." #not stranded
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	128 phase = "." #not phased
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	129 tags = "Note=%s" % cut_method
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	130
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	131 tab_handle = open(tabular_file)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	132 line = tab_handle.readline()
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	133 assert line.startswith("#ID\t"), line
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	134
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	135 gff_handle = open(gff_file, "w")
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	136 gff_handle.write("##gff-version 3\n")
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	137
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	138 for (title, seq), line in zip(fasta_iterator(fasta_file), tab_handle):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	139 parts = line.rstrip("\n").split("\t")
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	140 seqid = parts[0]
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	141 assert title.startswith(seqid), "%s vs %s" % (seqid, title)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	142 if len(seq)==0:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	143 #Is it possible to have a zero length reference in GFF3?
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	144 continue
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	145 cut = int(parts[cut_col])
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	146 if cut == 0:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	147 assert cut_method == "HMM_Cmax", cut_method
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	148 #TODO - Why does it do this?
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	149 cut = 1
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	150 assert 1 <= cut <= len(seq), "%i for %s len %i" % (cut, seqid, len(seq))
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	151 score = parts[score_col]
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	152 gff_handle.write("##sequence-region %s %i %i\n" \
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	153 % (seqid, 1, len(seq)))
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	154 #If the cut is at the very begining, there is no signal peptide!
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	155 if cut > 1:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	156 #signal_peptide = SO:0000418
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	157 gff_handle.write("%s\t%s\t%s\t%i\t%i\t%s\t%s\t%s\t%s\n" \
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	158 % (seqid, source,
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	159 "signal_peptide", 1, cut-1,
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	160 score, strand, phase, tags))
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	161 #mature_protein_region = SO:0000419
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	162 gff_handle.write("%s\t%s\t%s\t%i\t%i\t%s\t%s\t%s\t%s\n" \
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	163 % (seqid, source,
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	164 "mature_protein_region", cut, len(seq),
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	165 score, strand, phase, tags))
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	166 tab_handle.close()
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	167 gff_handle.close()
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	168
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	169
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	170 fasta_files = split_fasta(fasta_file, os.path.join(tmp_dir, "signalp"),
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	171 n=FASTA_CHUNK, truncate=truncate, max_len=MAX_LEN)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	172 temp_files = [f+".out" for f in fasta_files]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	173 assert len(fasta_files) == len(temp_files)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	174 jobs = ["signalp -short -t %s %s > %s" % (organism, fasta, temp)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	175 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	176 assert len(fasta_files) == len(temp_files) == len(jobs)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	177
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	178 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	179 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	180 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	181 os.remove(f)
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	182 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	183 os.rmdir(tmp_dir)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	184 except:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	185 pass
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	186
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	187 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	188 #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	189 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	190 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	191 assert len(fasta_files) == len(temp_files) == len(jobs)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	192 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	193 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	194 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	195 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	196 except IOError:
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	197 output = "(no output)"
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	198 if error_level or output.lower().startswith("error running"):
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	199 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	200 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	201 error_level)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	202 del results
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	203
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	204 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	205 fields = ["ID"]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	206 #NN results:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	207 for name in ["Cmax", "Ymax", "Smax"]:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	208 fields.extend(["NN_%s_score"%name, "NN_%s_pos"%name, "NN_%s_pred"%name])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	209 fields.extend(["NN_Smean_score", "NN_Smean_pred", "NN_D_score", "NN_D_pred"])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	210 #HMM results:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	211 fields.extend(["HMM_type", "HMM_Cmax_score", "HMM_Cmax_pos", "HMM_Cmax_pred",
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	212 "HMM_Sprob_score", "HMM_Sprob_pred"])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	213 out_handle.write("#" + "\t".join(fields) + "\n")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	214 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	215 data_handle = open(temp)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	216 clean_tabular(data_handle, out_handle)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	217 data_handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	218 out_handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	219
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	220 #GFF3:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	221 if cut_method:
8 391a142c1e60 Uploaded peterjc parents: 7 diff changeset	222 make_gff(fasta_file, tabular_file, gff3_file, cut_method)
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	223
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 5 diff changeset	224 clean_up(fasta_files + temp_files)

Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/signalp3.py @ 16:1d5486a02bc4 draft