line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
# Copyright Balamurugan Kumarasamy |
2
|
|
|
|
|
|
|
# You may distribute this module under the same terms as perl itself |
3
|
|
|
|
|
|
|
# POD documentation - main docs before the code |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
=head1 NAME |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
Bio::Tools::Run::Seg - Object for identifying low complexity |
8
|
|
|
|
|
|
|
regions in a given protein seequence. |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
=head1 SYNOPSIS |
11
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
# Build a Seg factory |
13
|
|
|
|
|
|
|
# $paramfile is the full path to the seg binary file |
14
|
|
|
|
|
|
|
my @params = ('PROGRAM',$paramfile); |
15
|
|
|
|
|
|
|
my $factory = Bio::Tools::Run::Seg->new($param); |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
# Pass the factory a Bio::Seq object |
18
|
|
|
|
|
|
|
# @feats is an array of Bio::SeqFeature::Generic objects |
19
|
|
|
|
|
|
|
my @feats = $factory->run($seq); |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
=head1 DESCRIPTION |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
Seg is a program which identifies low complexity regions in proteins. |
24
|
|
|
|
|
|
|
It was developed by Wootton and Federhen at NCBI. |
25
|
|
|
|
|
|
|
|
26
|
|
|
|
|
|
|
=head1 FEEDBACK |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
=head2 Mailing Lists |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
User feedback is an integral part of the evolution of this and other |
31
|
|
|
|
|
|
|
Bioperl modules. Send your comments and suggestions preferably to one |
32
|
|
|
|
|
|
|
of the Bioperl mailing lists. Your participation is much appreciated. |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
bioperl-l@bioperl.org - General discussion |
35
|
|
|
|
|
|
|
http://bioperl.org/wiki/Mailing_lists - About the mailing lists |
36
|
|
|
|
|
|
|
|
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 - Bala |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
Email savikalpa@fugu-sg.org |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
=head1 APPENDIX |
62
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
The rest of the documentation details each of the object |
64
|
|
|
|
|
|
|
methods. Internal methods are usually preceded with a _ |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
=cut |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
package Bio::Tools::Run::Seg; |
69
|
|
|
|
|
|
|
|
70
|
1
|
|
|
|
|
61
|
use vars qw($AUTOLOAD @ISA $PROGRAM $PROGRAMDIR |
71
|
1
|
|
|
1
|
|
99880
|
$PROGRAMNAME @SEG_PARAMS %OK_FIELD); |
|
1
|
|
|
|
|
1
|
|
72
|
1
|
|
|
1
|
|
4
|
use strict; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
14
|
|
73
|
1
|
|
|
1
|
|
452
|
use Bio::SeqIO; |
|
1
|
|
|
|
|
37842
|
|
|
1
|
|
|
|
|
22
|
|
74
|
1
|
|
|
1
|
|
6
|
use Bio::Root::Root; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
12
|
|
75
|
1
|
|
|
1
|
|
3
|
use Bio::Root::IO; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
12
|
|
76
|
1
|
|
|
1
|
|
396
|
use Bio::Tools::Seg; |
|
1
|
|
|
|
|
40198
|
|
|
1
|
|
|
|
|
21
|
|
77
|
1
|
|
|
1
|
|
481
|
use Bio::Tools::Run::WrapperBase; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
46
|
|
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
@ISA = qw(Bio::Root::Root Bio::Tools::Run::WrapperBase); |
80
|
|
|
|
|
|
|
|
81
|
|
|
|
|
|
|
BEGIN { |
82
|
1
|
|
|
1
|
|
3
|
@SEG_PARAMS=qw(PROGRAM VERBOSE); |
83
|
1
|
|
|
|
|
2
|
foreach my $attr ( @SEG_PARAMS) |
84
|
2
|
|
|
|
|
502
|
{ $OK_FIELD{$attr}++; } |
85
|
|
|
|
|
|
|
} |
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
=head2 program_name |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
Title : program_name |
90
|
|
|
|
|
|
|
Usage : $factory->program_name() |
91
|
|
|
|
|
|
|
Function: holds the program name |
92
|
|
|
|
|
|
|
Returns: string |
93
|
|
|
|
|
|
|
Args : None |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=cut |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
sub program_name { |
98
|
6
|
|
|
6
|
1
|
23
|
return 'seg'; |
99
|
|
|
|
|
|
|
} |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
=head2 program_dir |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
Title : program_dir |
104
|
|
|
|
|
|
|
Usage : $factory->program_dir(@params) |
105
|
|
|
|
|
|
|
Function: returns the program directory, obtained from ENV variable. |
106
|
|
|
|
|
|
|
Returns : string, or undef if $SEGDIR not in ENV |
107
|
|
|
|
|
|
|
Args : None |
108
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
=cut |
110
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
sub program_dir { |
112
|
3
|
50
|
|
3
|
1
|
10
|
return Bio::Root::IO->catfile($ENV{SEGDIR}) if $ENV{SEGDIR}; |
113
|
|
|
|
|
|
|
} |
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
sub AUTOLOAD { |
116
|
0
|
|
|
0
|
|
0
|
my $self = shift; |
117
|
0
|
|
|
|
|
0
|
my $attr = $AUTOLOAD; |
118
|
0
|
|
|
|
|
0
|
$attr =~ s/.*:://; |
119
|
0
|
|
|
|
|
0
|
$attr = uc $attr; |
120
|
0
|
0
|
|
|
|
0
|
$self->throw("Unallowed parameter: $attr !") unless $OK_FIELD{$attr}; |
121
|
0
|
0
|
|
|
|
0
|
$self->{$attr} = shift if @_; |
122
|
0
|
|
|
|
|
0
|
return $self->{$attr}; |
123
|
|
|
|
|
|
|
} |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
=head2 new |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
Title : new |
128
|
|
|
|
|
|
|
Usage : $rm->new(@params) |
129
|
|
|
|
|
|
|
Function: creates a new Seg factory |
130
|
|
|
|
|
|
|
Returns: Bio::Tools::Run::Seg |
131
|
|
|
|
|
|
|
Args : |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
=cut |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
sub new { |
136
|
1
|
|
|
1
|
1
|
82
|
my ($class,@args) = @_; |
137
|
1
|
|
|
|
|
11
|
my $self = $class->SUPER::new(@args); |
138
|
|
|
|
|
|
|
|
139
|
1
|
|
|
|
|
9
|
my ($attr, $value); |
140
|
1
|
|
|
|
|
4
|
while (@args) { |
141
|
0
|
|
|
|
|
0
|
$attr = shift @args; |
142
|
0
|
|
|
|
|
0
|
$value = shift @args; |
143
|
0
|
0
|
|
|
|
0
|
next if( $attr =~ /^-/ ); |
144
|
0
|
|
|
|
|
0
|
$self->$attr($value); |
145
|
|
|
|
|
|
|
} |
146
|
1
|
|
|
|
|
2
|
return $self; |
147
|
|
|
|
|
|
|
} |
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
=head2 predict_protein_features |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
Title : predict_protein_features() |
152
|
|
|
|
|
|
|
Usage : DEPRECATED Use $obj->run($seq) instead |
153
|
|
|
|
|
|
|
Function: Runs Seg and creates an array of featrues |
154
|
|
|
|
|
|
|
Returns : An array of Bio::SeqFeature::Generic objects |
155
|
|
|
|
|
|
|
Args : A Bio::PrimarySeqI |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
=cut |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
sub predict_protein_features{ |
160
|
0
|
|
|
0
|
1
|
|
return shift->run(@_); |
161
|
|
|
|
|
|
|
} |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
=head2 run |
164
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
Title : run |
166
|
|
|
|
|
|
|
Usage : $obj->run($seq) |
167
|
|
|
|
|
|
|
Function: Runs Seg and creates an array of featrues |
168
|
|
|
|
|
|
|
Returns : An array of Bio::SeqFeature::Generic objects |
169
|
|
|
|
|
|
|
Args : A Bio::PrimarySeqI |
170
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
=cut |
172
|
|
|
|
|
|
|
|
173
|
|
|
|
|
|
|
sub run{ |
174
|
0
|
|
|
0
|
1
|
|
my ($self,$seq) = @_; |
175
|
0
|
|
|
|
|
|
my @feats; |
176
|
|
|
|
|
|
|
|
177
|
0
|
0
|
|
|
|
|
if (ref($seq) ) { # it is an object |
178
|
0
|
0
|
|
|
|
|
if (ref($seq) =~ /GLOB/) { |
179
|
0
|
|
|
|
|
|
$self->throw("cannot use filehandle"); |
180
|
|
|
|
|
|
|
} |
181
|
|
|
|
|
|
|
|
182
|
0
|
|
|
|
|
|
my $infile1 = $self->_writeSeqFile($seq); |
183
|
|
|
|
|
|
|
|
184
|
0
|
|
|
|
|
|
$self->_input($infile1); |
185
|
|
|
|
|
|
|
|
186
|
0
|
|
|
|
|
|
@feats = $self->_run(); |
187
|
0
|
|
|
|
|
|
unlink $infile1; |
188
|
|
|
|
|
|
|
} |
189
|
|
|
|
|
|
|
else { |
190
|
|
|
|
|
|
|
#The seq object is not a seq object but a file. |
191
|
|
|
|
|
|
|
#Here the file does not need to be created. |
192
|
|
|
|
|
|
|
|
193
|
0
|
|
|
|
|
|
$self->_input($seq); |
194
|
|
|
|
|
|
|
|
195
|
0
|
|
|
|
|
|
@feats = $self->_run(); |
196
|
|
|
|
|
|
|
|
197
|
|
|
|
|
|
|
} |
198
|
|
|
|
|
|
|
|
199
|
0
|
|
|
|
|
|
return @feats; |
200
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
} |
202
|
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
=head2 _input |
204
|
|
|
|
|
|
|
|
205
|
|
|
|
|
|
|
Title : _input |
206
|
|
|
|
|
|
|
Usage : obj->_input($seqFile) |
207
|
|
|
|
|
|
|
Function: Internal (not to be used directly) |
208
|
|
|
|
|
|
|
Returns : |
209
|
|
|
|
|
|
|
Args : |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
=cut |
212
|
|
|
|
|
|
|
|
213
|
|
|
|
|
|
|
sub _input { |
214
|
0
|
|
|
0
|
|
|
my ($self,$infile1) = @_; |
215
|
0
|
0
|
|
|
|
|
if(defined $infile1){ |
216
|
|
|
|
|
|
|
|
217
|
0
|
|
|
|
|
|
$self->{'input'}=$infile1; |
218
|
|
|
|
|
|
|
} |
219
|
0
|
|
|
|
|
|
return $self->{'input'}; |
220
|
|
|
|
|
|
|
} |
221
|
|
|
|
|
|
|
|
222
|
|
|
|
|
|
|
=head2 _run |
223
|
|
|
|
|
|
|
|
224
|
|
|
|
|
|
|
Title : _run |
225
|
|
|
|
|
|
|
Usage : $obj->_run() |
226
|
|
|
|
|
|
|
Function: Internal (not to be used directly) |
227
|
|
|
|
|
|
|
Returns : An array of Bio::SeqFeature::Generic objects |
228
|
|
|
|
|
|
|
Args : None |
229
|
|
|
|
|
|
|
|
230
|
|
|
|
|
|
|
=cut |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
sub _run { |
233
|
0
|
|
|
0
|
|
|
my ($self)= @_; |
234
|
|
|
|
|
|
|
|
235
|
0
|
|
|
|
|
|
my ($tfh1,$outfile) = $self->io->tempfile(-dir=>$self->tempdir()); |
236
|
0
|
|
|
|
|
|
my $str =$self->executable." ".$self->_input." -l > ".$outfile; |
237
|
0
|
|
|
|
|
|
my $status = system($str); |
238
|
0
|
0
|
|
|
|
|
$self->throw( "Seg call ($str) crashed: $? \n") unless $status==0; |
239
|
|
|
|
|
|
|
|
240
|
0
|
|
|
|
|
|
my $filehandle; |
241
|
0
|
0
|
|
|
|
|
if (ref ($outfile) !~ /GLOB/) { |
242
|
0
|
0
|
|
|
|
|
open (SEG, "<".$outfile) or $self->throw ("Couldn't open file ".$outfile.": $!\n"); |
243
|
0
|
|
|
|
|
|
$filehandle = \*SEG; |
244
|
|
|
|
|
|
|
} |
245
|
|
|
|
|
|
|
else { |
246
|
0
|
|
|
|
|
|
$filehandle = $outfile; |
247
|
|
|
|
|
|
|
} |
248
|
0
|
|
|
|
|
|
my $seg_parser = Bio::Tools::Seg->new(-fh=>$filehandle); |
249
|
|
|
|
|
|
|
|
250
|
0
|
|
|
|
|
|
my @seg_feat; |
251
|
|
|
|
|
|
|
|
252
|
0
|
|
|
|
|
|
while(my $seg_feat = $seg_parser->next_result){ |
253
|
|
|
|
|
|
|
|
254
|
0
|
|
|
|
|
|
push @seg_feat, $seg_feat; |
255
|
|
|
|
|
|
|
} |
256
|
|
|
|
|
|
|
# free resources |
257
|
0
|
|
|
|
|
|
$self->cleanup(); |
258
|
0
|
|
|
|
|
|
unlink $outfile; |
259
|
0
|
|
|
|
|
|
close($tfh1); |
260
|
0
|
|
|
|
|
|
undef $tfh1; |
261
|
0
|
|
|
|
|
|
return @seg_feat; |
262
|
|
|
|
|
|
|
|
263
|
|
|
|
|
|
|
} |
264
|
|
|
|
|
|
|
|
265
|
|
|
|
|
|
|
=head2 _writeSeqFile |
266
|
|
|
|
|
|
|
|
267
|
|
|
|
|
|
|
Title : _writeSeqFile |
268
|
|
|
|
|
|
|
Usage : obj->_writeSeqFile($seq) |
269
|
|
|
|
|
|
|
Function: Internal (not to be used directly) |
270
|
|
|
|
|
|
|
Returns : string - Fasta filename to which $seq was written |
271
|
|
|
|
|
|
|
Args : Bio::Seq object |
272
|
|
|
|
|
|
|
|
273
|
|
|
|
|
|
|
=cut |
274
|
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
sub _writeSeqFile{ |
276
|
0
|
|
|
0
|
|
|
my ($self,$seq) = @_; |
277
|
0
|
|
|
|
|
|
my ($tfh,$inputfile) = $self->io->tempfile(-dir=>$self->tempdir()); |
278
|
0
|
|
|
|
|
|
my $in = Bio::SeqIO->new(-fh => $tfh , '-format' => 'Fasta'); |
279
|
0
|
|
|
|
|
|
$in->write_seq($seq); |
280
|
0
|
|
|
|
|
|
$in->close(); |
281
|
0
|
|
|
|
|
|
close($tfh); |
282
|
0
|
|
|
|
|
|
undef $tfh; |
283
|
0
|
|
|
|
|
|
return $inputfile; |
284
|
|
|
|
|
|
|
|
285
|
|
|
|
|
|
|
} |
286
|
|
|
|
|
|
|
1; |