line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
# Wrapper module for SignalP Bio::Tools::Run::Signalp |
2
|
|
|
|
|
|
|
# |
3
|
|
|
|
|
|
|
# Based on the EnsEMBL module Bio::EnsEMBL::Pipeline::Runnable::Protein::Signalp |
4
|
|
|
|
|
|
|
# originally written by Marc Sohrmann (ms2@sanger.ac.uk) |
5
|
|
|
|
|
|
|
# Written in BioPipe by Balamurugan Kumarasamy |
6
|
|
|
|
|
|
|
# Please direct questions and support issues to |
7
|
|
|
|
|
|
|
# |
8
|
|
|
|
|
|
|
# Cared for by the Fugu Informatics team (fuguteam@fugu-sg.org) |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
=head1 NAME |
11
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
Bio::Tools::Run::Signalp |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
=head1 SYNOPSIS |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
Build a Signalp factory |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
my $factory = Bio::Tools::Run::Signalp->new(); |
19
|
|
|
|
|
|
|
# Pass the factory a Bio::Seq object |
20
|
|
|
|
|
|
|
# @feats is an array of Bio::SeqFeature::Generic objects |
21
|
|
|
|
|
|
|
my @feats = $factory->run($seq); |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
=head1 DESCRIPTION |
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
wrapper module for Signalp program |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
=head1 FEEDBACK |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
=head2 Mailing Lists |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
User feedback is an integral part of the evolution of this and other |
32
|
|
|
|
|
|
|
Bioperl modules. Send your comments and suggestions preferably to one |
33
|
|
|
|
|
|
|
of the Bioperl mailing lists. Your participation is much appreciated. |
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
bioperl-l@bioperl.org - General discussion |
36
|
|
|
|
|
|
|
http://bioperl.org/wiki/Mailing_lists - About the mailing lists |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
=head2 Support |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
Please direct usage questions or support issues to the mailing list: |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
I |
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
rather than to the module maintainer directly. Many experienced and |
45
|
|
|
|
|
|
|
reponsive experts will be able look at the problem and quickly |
46
|
|
|
|
|
|
|
address it. Please include a thorough description of the problem |
47
|
|
|
|
|
|
|
with code and data examples if at all possible. |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
=head2 Reporting Bugs |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
Report bugs to the Bioperl bug tracking system to help us keep track |
52
|
|
|
|
|
|
|
the bugs and their resolution. Bug reports can be submitted via the |
53
|
|
|
|
|
|
|
web: |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
http://redmine.open-bio.org/projects/bioperl/ |
56
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
=head1 AUTHOR |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
Based on the EnsEMBL module Bio::EnsEMBL::Pipeline::Runnable::Protein::Signalp |
60
|
|
|
|
|
|
|
originally written by Marc Sohrmann (ms2@sanger.ac.uk) |
61
|
|
|
|
|
|
|
Written in BioPipe by Balamurugan Kumarasamy |
62
|
|
|
|
|
|
|
Contributions by David Vilanova (david.vilanova@urbanet.ch) |
63
|
|
|
|
|
|
|
Shawn Hoon (shawnh@fugu-sg.org) |
64
|
|
|
|
|
|
|
# Please direct questions and support issues to |
65
|
|
|
|
|
|
|
# |
66
|
|
|
|
|
|
|
Cared for by the Fugu Informatics team (fuguteam@fugu-sg.org) |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
=head1 APPENDIX |
69
|
|
|
|
|
|
|
|
70
|
|
|
|
|
|
|
The rest of the documentation details each of the object |
71
|
|
|
|
|
|
|
methods. Internal methods are usually preceded with a _ |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
=cut |
74
|
|
|
|
|
|
|
|
75
|
|
|
|
|
|
|
package Bio::Tools::Run::Signalp; |
76
|
|
|
|
|
|
|
|
77
|
1
|
|
|
|
|
69
|
use vars qw($AUTOLOAD @ISA $PROGRAM $PROGRAMDIR |
78
|
1
|
|
|
1
|
|
105393
|
$PROGRAMNAME @SIGNALP_PARAMS %OK_FIELD); |
|
1
|
|
|
|
|
2
|
|
79
|
1
|
|
|
1
|
|
4
|
use strict; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
15
|
|
80
|
1
|
|
|
1
|
|
476
|
use Bio::SeqIO; |
|
1
|
|
|
|
|
41039
|
|
|
1
|
|
|
|
|
28
|
|
81
|
1
|
|
|
1
|
|
7
|
use Bio::Root::Root; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
13
|
|
82
|
1
|
|
|
1
|
|
3
|
use Bio::Root::IO; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
13
|
|
83
|
1
|
|
|
1
|
|
427
|
use Bio::Factory::ApplicationFactoryI; |
|
1
|
|
|
|
|
101
|
|
|
1
|
|
|
|
|
18
|
|
84
|
1
|
|
|
1
|
|
429
|
use Bio::Tools::Signalp; |
|
1
|
|
|
|
|
42673
|
|
|
1
|
|
|
|
|
27
|
|
85
|
1
|
|
|
1
|
|
456
|
use Bio::Tools::Run::WrapperBase; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
51
|
|
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
@ISA = qw(Bio::Root::Root Bio::Tools::Run::WrapperBase); |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
BEGIN { |
90
|
1
|
|
|
1
|
|
3
|
@SIGNALP_PARAMS=qw(PROGRAM VERBOSE); |
91
|
1
|
|
|
|
|
2
|
foreach my $attr ( @SIGNALP_PARAMS) |
92
|
2
|
|
|
|
|
568
|
{ $OK_FIELD{$attr}++; } |
93
|
|
|
|
|
|
|
} |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=head2 program_name |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
Title : program_name |
98
|
|
|
|
|
|
|
Usage : $factory>program_name() |
99
|
|
|
|
|
|
|
Function: holds the program name |
100
|
|
|
|
|
|
|
Returns: string |
101
|
|
|
|
|
|
|
Args : None |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
=cut |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
sub program_name { |
106
|
6
|
|
|
6
|
1
|
23
|
return 'signalp'; |
107
|
|
|
|
|
|
|
} |
108
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
=head2 program_dir |
110
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
Title : program_dir |
112
|
|
|
|
|
|
|
Usage : $factory->program_dir(@params) |
113
|
|
|
|
|
|
|
Function: returns the program directory, obtained from ENV variable. |
114
|
|
|
|
|
|
|
Returns: string |
115
|
|
|
|
|
|
|
Args : |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
=cut |
118
|
|
|
|
|
|
|
|
119
|
|
|
|
|
|
|
sub program_dir { |
120
|
3
|
50
|
|
3
|
1
|
14
|
return Bio::Root::IO->catfile($ENV{SIGNALPDIR}) if $ENV{SIGNALPDIR}; |
121
|
|
|
|
|
|
|
} |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
sub AUTOLOAD { |
124
|
0
|
|
|
0
|
|
0
|
my $self = shift; |
125
|
0
|
|
|
|
|
0
|
my $attr = $AUTOLOAD; |
126
|
0
|
0
|
|
|
|
0
|
return $self->$attr if $self->$attr; |
127
|
0
|
|
|
|
|
0
|
$attr =~ s/.*:://; |
128
|
0
|
|
|
|
|
0
|
$attr = uc $attr; |
129
|
0
|
0
|
|
|
|
0
|
$self->throw("Unallowed parameter: $attr !") unless $OK_FIELD{$attr}; |
130
|
0
|
0
|
|
|
|
0
|
$self->{$attr} = shift if @_; |
131
|
0
|
|
|
|
|
0
|
return $self->{$attr}; |
132
|
|
|
|
|
|
|
} |
133
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
=head2 new |
135
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
Title : new |
137
|
|
|
|
|
|
|
Usage : my $factory= Bio::Tools::Run::Signalp->new(); |
138
|
|
|
|
|
|
|
Function: creates a new Signalp factory |
139
|
|
|
|
|
|
|
Returns: Bio::Tools::Run::Signalp |
140
|
|
|
|
|
|
|
Args : |
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
=cut |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
sub new { |
145
|
1
|
|
|
1
|
1
|
90
|
my ($class,@args) = @_; |
146
|
1
|
|
|
|
|
12
|
my $self = $class->SUPER::new(@args); |
147
|
|
|
|
|
|
|
|
148
|
1
|
|
|
|
|
9
|
my ($attr, $value); |
149
|
1
|
|
|
|
|
4
|
while (@args) { |
150
|
0
|
|
|
|
|
0
|
$attr = shift @args; |
151
|
0
|
|
|
|
|
0
|
$value = shift @args; |
152
|
0
|
0
|
|
|
|
0
|
next if( $attr =~ /^-/ ); # don't want named parameters |
153
|
0
|
0
|
|
|
|
0
|
if ($attr =~/PROGRAM/i) { |
154
|
0
|
|
|
|
|
0
|
$self->executable($value); |
155
|
0
|
|
|
|
|
0
|
next; |
156
|
|
|
|
|
|
|
} |
157
|
0
|
|
|
|
|
0
|
$self->$attr($value); |
158
|
|
|
|
|
|
|
} |
159
|
1
|
|
|
|
|
3
|
return $self; |
160
|
|
|
|
|
|
|
} |
161
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
=head2 predict_protein_features |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
Title : predict_protein_features() |
165
|
|
|
|
|
|
|
Usage : DEPRECATED. Use $factory->run($seq) instead |
166
|
|
|
|
|
|
|
Function: Runs Signalp and creates an array of featrues |
167
|
|
|
|
|
|
|
Returns : An array of Bio::SeqFeature::Generic objects |
168
|
|
|
|
|
|
|
Args : A Bio::PrimarySeqI |
169
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
=cut |
171
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
sub predict_protein_features{ |
173
|
0
|
|
|
0
|
1
|
|
return shift->run(@_); |
174
|
|
|
|
|
|
|
} |
175
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
=head2 run |
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
Title : run() |
179
|
|
|
|
|
|
|
Usage : my $feats = $factory->run($seq) |
180
|
|
|
|
|
|
|
Function: Runs Signalp |
181
|
|
|
|
|
|
|
Returns : An array of Bio::SeqFeature::Generic objects |
182
|
|
|
|
|
|
|
Args : A Bio::PrimarySeqI |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
=cut |
185
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
sub run { |
187
|
0
|
|
|
0
|
1
|
|
my ($self,$seq) = @_; |
188
|
0
|
|
|
|
|
|
my @feats; |
189
|
|
|
|
|
|
|
|
190
|
0
|
0
|
|
|
|
|
if (ref($seq) ) { |
191
|
|
|
|
|
|
|
|
192
|
0
|
0
|
|
|
|
|
if (ref($seq) =~ /GLOB/) { |
193
|
0
|
|
|
|
|
|
$self->throw("cannot use filehandle"); |
194
|
|
|
|
|
|
|
} |
195
|
|
|
|
|
|
|
|
196
|
0
|
|
|
|
|
|
my $infile1 = $self->_writeSeqFile($seq); |
197
|
|
|
|
|
|
|
|
198
|
0
|
|
|
|
|
|
$self->_input($infile1); |
199
|
|
|
|
|
|
|
|
200
|
0
|
|
|
|
|
|
@feats = $self->_run(); |
201
|
0
|
|
|
|
|
|
unlink $infile1; |
202
|
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
} |
204
|
|
|
|
|
|
|
else { |
205
|
0
|
|
|
|
|
|
my $in = Bio::SeqIO->new(-file => $seq, '-format' =>'fasta'); |
206
|
0
|
|
|
|
|
|
my $infile1; |
207
|
|
|
|
|
|
|
|
208
|
0
|
|
|
|
|
|
while ( my $tmpseq = $in->next_seq() ) { |
209
|
0
|
|
|
|
|
|
$infile1 = $self->_writeSeqFile($tmpseq); |
210
|
|
|
|
|
|
|
} |
211
|
|
|
|
|
|
|
|
212
|
0
|
|
|
|
|
|
$self->_input($infile1); |
213
|
|
|
|
|
|
|
|
214
|
0
|
|
|
|
|
|
@feats = $self->_run(); |
215
|
|
|
|
|
|
|
} |
216
|
|
|
|
|
|
|
|
217
|
0
|
|
|
|
|
|
return @feats; |
218
|
|
|
|
|
|
|
} |
219
|
|
|
|
|
|
|
|
220
|
|
|
|
|
|
|
=head2 _input |
221
|
|
|
|
|
|
|
|
222
|
|
|
|
|
|
|
Title : _input |
223
|
|
|
|
|
|
|
Usage : $factory->_input($seqFile) |
224
|
|
|
|
|
|
|
Function: get/set for input file |
225
|
|
|
|
|
|
|
Returns : |
226
|
|
|
|
|
|
|
Args : |
227
|
|
|
|
|
|
|
|
228
|
|
|
|
|
|
|
=cut |
229
|
|
|
|
|
|
|
|
230
|
|
|
|
|
|
|
sub _input() { |
231
|
0
|
|
|
0
|
|
|
my ($self,$infile1) = @_; |
232
|
0
|
0
|
|
|
|
|
$self->{'input'} = $infile1 if(defined $infile1); |
233
|
0
|
|
|
|
|
|
return $self->{'input'}; |
234
|
|
|
|
|
|
|
} |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
=head2 _run |
237
|
|
|
|
|
|
|
|
238
|
|
|
|
|
|
|
Title : _run |
239
|
|
|
|
|
|
|
Usage : $factory->_run() |
240
|
|
|
|
|
|
|
Function: Makes a system call and runs signalp |
241
|
|
|
|
|
|
|
Returns : An array of Bio::SeqFeature::Generic objects |
242
|
|
|
|
|
|
|
Args : |
243
|
|
|
|
|
|
|
|
244
|
|
|
|
|
|
|
=cut |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
sub _run { |
247
|
0
|
|
|
0
|
|
|
my ($self)= @_; |
248
|
|
|
|
|
|
|
|
249
|
0
|
|
|
|
|
|
my ($tfh1,$outfile) = $self->io->tempfile(-dir=>$self->tempdir()); |
250
|
0
|
|
|
|
|
|
my $str =$self->executable." -t euk -trunc 50 ".$self->{'input'}." > ".$outfile; |
251
|
0
|
|
|
|
|
|
my $status = system($str); |
252
|
0
|
0
|
|
|
|
|
$self->throw( "Signalp call ($str) crashed: $? \n") unless $status==0; |
253
|
|
|
|
|
|
|
|
254
|
0
|
|
|
|
|
|
my $filehandle; |
255
|
0
|
0
|
|
|
|
|
if (ref ($outfile) !~ /GLOB/) { |
256
|
0
|
0
|
|
|
|
|
open (SIGNALP, "<".$outfile) or $self->throw ("Couldn't open file ".$outfile.": $!\n"); |
257
|
0
|
|
|
|
|
|
$filehandle = \*SIGNALP; |
258
|
|
|
|
|
|
|
} |
259
|
|
|
|
|
|
|
else { |
260
|
0
|
|
|
|
|
|
$filehandle = $outfile; |
261
|
|
|
|
|
|
|
} |
262
|
|
|
|
|
|
|
|
263
|
0
|
|
|
|
|
|
my $signalp_parser = Bio::Tools::Signalp->new(-fh=>$filehandle); |
264
|
|
|
|
|
|
|
|
265
|
0
|
|
|
|
|
|
my @signalp_feat; |
266
|
|
|
|
|
|
|
|
267
|
0
|
|
|
|
|
|
while(my $signalp_feat = $signalp_parser->next_result){ |
268
|
|
|
|
|
|
|
|
269
|
0
|
|
|
|
|
|
push @signalp_feat, $signalp_feat; |
270
|
|
|
|
|
|
|
} |
271
|
|
|
|
|
|
|
|
272
|
0
|
|
|
|
|
|
$self->cleanup(); |
273
|
0
|
|
|
|
|
|
close($tfh1); |
274
|
0
|
|
|
|
|
|
undef $tfh1; |
275
|
0
|
|
|
|
|
|
unlink $outfile; |
276
|
|
|
|
|
|
|
|
277
|
0
|
|
|
|
|
|
return @signalp_feat; |
278
|
|
|
|
|
|
|
} |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
|
281
|
|
|
|
|
|
|
=head2 _writeSeqFile |
282
|
|
|
|
|
|
|
|
283
|
|
|
|
|
|
|
Title : _writeSeqFile |
284
|
|
|
|
|
|
|
Usage : $factory->_writeSeqFile($seq) |
285
|
|
|
|
|
|
|
Function: Creates a file from the given seq object |
286
|
|
|
|
|
|
|
Returns : A string(filename) |
287
|
|
|
|
|
|
|
Args : Bio::PrimarySeqI |
288
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
=cut |
290
|
|
|
|
|
|
|
|
291
|
|
|
|
|
|
|
sub _writeSeqFile{ |
292
|
0
|
|
|
0
|
|
|
my ($self,$seq) = @_; |
293
|
0
|
|
|
|
|
|
my ($tfh,$inputfile) = $self->io->tempfile(-dir=>$self->tempdir()); |
294
|
0
|
|
|
|
|
|
my $in = Bio::SeqIO->new(-fh => $tfh , '-format' => 'fasta'); |
295
|
0
|
|
|
|
|
|
$in->write_seq($seq); |
296
|
0
|
|
|
|
|
|
$in->close(); |
297
|
0
|
|
|
|
|
|
close($tfh); |
298
|
0
|
|
|
|
|
|
undef $tfh; |
299
|
0
|
|
|
|
|
|
return $inputfile; |
300
|
|
|
|
|
|
|
|
301
|
|
|
|
|
|
|
} |
302
|
|
|
|
|
|
|
1; |