line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
#--------------------------------------------------------- |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
=head1 NAME |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
Bio::Matrix::PSM::PsmHeaderI - handles the header data from a PSM file |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
=head1 SYNOPSIS |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
use Bio::Matrix::PSM::IO; |
10
|
|
|
|
|
|
|
#Obtain an Bio::Matrix::PSM::IO object: |
11
|
|
|
|
|
|
|
my $psmIO= Bio::Matrix::PSM::IO->new(-file=>$file, -format=>'mast'); |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
#Get some general data about the file you are parsing: |
14
|
|
|
|
|
|
|
my $release=$psmIO->release; |
15
|
|
|
|
|
|
|
my $version=$psmIO->version; |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
print "This analysis was performed using MAST version $version, release $release\n"; |
18
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
#Now let's see what are the consensus sequences of the motifs fed as an input: |
20
|
|
|
|
|
|
|
my %seq=$psmIO->seq; |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
#let's cycle through all consensus sequences now: |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
foreach my $id ($psmIO->hid) { |
25
|
|
|
|
|
|
|
print "Motif $id is \t",$seq{$id},"\n"; |
26
|
|
|
|
|
|
|
} |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
#Finally look at the stuff we do not parse: |
29
|
|
|
|
|
|
|
my @inputfile=grep(/datafile/i,$psmIO->unstructured); |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
=head1 DESCRIPTION |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
Generally you should not use this object directly, you can access the |
34
|
|
|
|
|
|
|
information through a PSM driver (See Bio::Matrix::PSM::IO). It is |
35
|
|
|
|
|
|
|
handling the header data from a PSM file which may be very |
36
|
|
|
|
|
|
|
different. This means that some of the methods will return undef |
37
|
|
|
|
|
|
|
naturally, because this information is not present in the file which |
38
|
|
|
|
|
|
|
is parsed. Some important data might be left over in the unstructured |
39
|
|
|
|
|
|
|
part, and you might have to parse it yourself. I will try to |
40
|
|
|
|
|
|
|
'structure' this header more in the near future. |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
=head1 FEEDBACK |
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
=head2 Mailing Lists |
46
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
User feedback is an integral part of the evolution of this |
48
|
|
|
|
|
|
|
and other Bioperl modules. Send your comments and suggestions preferably |
49
|
|
|
|
|
|
|
to one of the Bioperl mailing lists. |
50
|
|
|
|
|
|
|
Your participation is much appreciated. |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
bioperl-l@bioperl.org - General discussion |
53
|
|
|
|
|
|
|
http://bioperl.org/wiki/Mailing_lists - About the mailing lists |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
=head2 Support |
56
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
Please direct usage questions or support issues to the mailing list: |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
I |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
rather than to the module maintainer directly. Many experienced and |
62
|
|
|
|
|
|
|
reponsive experts will be able look at the problem and quickly |
63
|
|
|
|
|
|
|
address it. Please include a thorough description of the problem |
64
|
|
|
|
|
|
|
with code and data examples if at all possible. |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
=head2 Reporting Bugs |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
Report bugs to the Bioperl bug tracking system to help us keep track |
69
|
|
|
|
|
|
|
the bugs and their resolution. Bug reports can be submitted via the |
70
|
|
|
|
|
|
|
web: |
71
|
|
|
|
|
|
|
|
72
|
|
|
|
|
|
|
https://github.com/bioperl/bioperl-live/issues |
73
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
=head1 AUTHOR - Stefan Kirov |
75
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
Email skirov@utk.edu |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
=head1 APPENDIX |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
=cut |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
# Let the code begin... |
84
|
|
|
|
|
|
|
package Bio::Matrix::PSM::PsmHeaderI; |
85
|
3
|
|
|
3
|
|
14
|
use Bio::Matrix::PSM::InstanceSite; |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
64
|
|
86
|
3
|
|
|
3
|
|
10
|
use Bio::Matrix::PSM::Psm; |
|
3
|
|
|
|
|
4
|
|
|
3
|
|
|
|
|
39
|
|
87
|
3
|
|
|
3
|
|
9
|
use Bio::Matrix::PSM::IO; |
|
3
|
|
|
|
|
2
|
|
|
3
|
|
|
|
|
62
|
|
88
|
3
|
|
|
3
|
|
9
|
use strict; |
|
3
|
|
|
|
|
3
|
|
|
3
|
|
|
|
|
62
|
|
89
|
3
|
|
|
3
|
|
9
|
use base qw(Bio::Matrix::PSM::PsmI); |
|
3
|
|
|
|
|
3
|
|
|
3
|
|
|
|
|
854
|
|
90
|
|
|
|
|
|
|
|
91
|
|
|
|
|
|
|
#Accessor methods, based on the driver |
92
|
|
|
|
|
|
|
@Bio::Matrix::PSM::PsmHeader::MASTHEADER=qw(html version release |
93
|
|
|
|
|
|
|
seq hid length instances |
94
|
|
|
|
|
|
|
unstructured); |
95
|
|
|
|
|
|
|
@Bio::Matrix::PSM::PsmHeader::MEMEHEADER=qw(html version release hid |
96
|
|
|
|
|
|
|
weight length unstructured); |
97
|
|
|
|
|
|
|
@Bio::Matrix::PSM::PsmHeader::TRANSFACHEADER=qw(unstructured version release); |
98
|
|
|
|
|
|
|
@Bio::Matrix::PSM::PsmHeader::ALLHEADER=qw(header release type version html |
99
|
|
|
|
|
|
|
release weight length hid |
100
|
|
|
|
|
|
|
seq instances unstructured); |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
=head2 new |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
Title : new |
105
|
|
|
|
|
|
|
Usage : my $header= Bio::Matrix::PSM::PsmHeader->new |
106
|
|
|
|
|
|
|
( -seq=>\%seq, -mid=>\%mid, -width=>\%width, |
107
|
|
|
|
|
|
|
-instances=>\%instances, -header=>\@header, -type=>'mast'); |
108
|
|
|
|
|
|
|
Function: Creates a new Bio::Matrix::PSM::PsmHeader object |
109
|
|
|
|
|
|
|
Throws : |
110
|
|
|
|
|
|
|
Example : |
111
|
|
|
|
|
|
|
Returns : Bio::Matrix::PSM::PsmHeaderI object |
112
|
|
|
|
|
|
|
Args : hash |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
=cut |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
=head2 seq |
118
|
|
|
|
|
|
|
|
119
|
|
|
|
|
|
|
Title : seq |
120
|
|
|
|
|
|
|
Usage : my %seq= $header->seq(); |
121
|
|
|
|
|
|
|
Function: Returns the sequence data as a hash, indexed by a |
122
|
|
|
|
|
|
|
sequence ID (motif id or accession number) |
123
|
|
|
|
|
|
|
In case the input data is a motif it would return the |
124
|
|
|
|
|
|
|
consenus seq for each of them (mast). |
125
|
|
|
|
|
|
|
Throws : |
126
|
|
|
|
|
|
|
Example : |
127
|
|
|
|
|
|
|
Returns : hash |
128
|
|
|
|
|
|
|
Args : |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
=cut |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
sub seq { |
134
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
135
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
136
|
|
|
|
|
|
|
} |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
=head2 hid |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
Title : hid |
142
|
|
|
|
|
|
|
Usage : my @ids= $header->hid(); |
143
|
|
|
|
|
|
|
Function: Returns array with the motif/instance ids |
144
|
|
|
|
|
|
|
Throws : |
145
|
|
|
|
|
|
|
Example : |
146
|
|
|
|
|
|
|
Returns : array |
147
|
|
|
|
|
|
|
Args : |
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
=cut |
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
sub hid { |
153
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
154
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
155
|
|
|
|
|
|
|
} |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
=head2 length |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
Title : length |
160
|
|
|
|
|
|
|
Usage : my %length= $header->length(); |
161
|
|
|
|
|
|
|
Function: Returns the length of the input sequence or motifs as a hash, indexed |
162
|
|
|
|
|
|
|
by a sequence ID (motif id or accession number) |
163
|
|
|
|
|
|
|
Throws : |
164
|
|
|
|
|
|
|
Example : |
165
|
|
|
|
|
|
|
Returns : hash |
166
|
|
|
|
|
|
|
Args : |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
=cut |
170
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
sub length { |
172
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
173
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
174
|
|
|
|
|
|
|
} |
175
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
=head2 instances |
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
Title : instances |
179
|
|
|
|
|
|
|
Usage : my %instances= $header->length(); |
180
|
|
|
|
|
|
|
Function: Returns the instance, used as a hash, indexed |
181
|
|
|
|
|
|
|
by a sequence ID (motif id or accession number) |
182
|
|
|
|
|
|
|
Throws : |
183
|
|
|
|
|
|
|
Example : |
184
|
|
|
|
|
|
|
Returns : hash of Bio::Matrix::PSM::InstanceSite objects |
185
|
|
|
|
|
|
|
Args : |
186
|
|
|
|
|
|
|
|
187
|
|
|
|
|
|
|
|
188
|
|
|
|
|
|
|
=cut |
189
|
|
|
|
|
|
|
|
190
|
|
|
|
|
|
|
sub instances { |
191
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
192
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
193
|
|
|
|
|
|
|
} |
194
|
|
|
|
|
|
|
|
195
|
|
|
|
|
|
|
=head2 weights |
196
|
|
|
|
|
|
|
|
197
|
|
|
|
|
|
|
Title : weights |
198
|
|
|
|
|
|
|
Usage : my %weights= $header->weights(); |
199
|
|
|
|
|
|
|
Function: Returns the weights of the input sequence as a hash, indexed |
200
|
|
|
|
|
|
|
by a sequence ID |
201
|
|
|
|
|
|
|
Throws : |
202
|
|
|
|
|
|
|
Example : |
203
|
|
|
|
|
|
|
Returns : hash |
204
|
|
|
|
|
|
|
Args : |
205
|
|
|
|
|
|
|
|
206
|
|
|
|
|
|
|
|
207
|
|
|
|
|
|
|
=cut |
208
|
|
|
|
|
|
|
|
209
|
|
|
|
|
|
|
sub weights { |
210
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
211
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
212
|
|
|
|
|
|
|
} |
213
|
|
|
|
|
|
|
|
214
|
|
|
|
|
|
|
|
215
|
|
|
|
|
|
|
=head2 unstuctured |
216
|
|
|
|
|
|
|
|
217
|
|
|
|
|
|
|
Title : unstuctured |
218
|
|
|
|
|
|
|
Usage : my @unstructured= $header->unstuctured(); |
219
|
|
|
|
|
|
|
Function: Returns the unstructured data in the header as an array, one line per |
220
|
|
|
|
|
|
|
array element, all control symbols are removed with \W |
221
|
|
|
|
|
|
|
Throws : |
222
|
|
|
|
|
|
|
Example : |
223
|
|
|
|
|
|
|
Returns : array |
224
|
|
|
|
|
|
|
Args : |
225
|
|
|
|
|
|
|
|
226
|
|
|
|
|
|
|
|
227
|
|
|
|
|
|
|
=cut |
228
|
|
|
|
|
|
|
|
229
|
|
|
|
|
|
|
sub unstructured { |
230
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
231
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
232
|
|
|
|
|
|
|
} |
233
|
|
|
|
|
|
|
|
234
|
|
|
|
|
|
|
=head2 version |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
Title : version |
237
|
|
|
|
|
|
|
Usage : my $version= $header->version; |
238
|
|
|
|
|
|
|
Function: Returns the version of the file being parsed if such exists |
239
|
|
|
|
|
|
|
Throws : |
240
|
|
|
|
|
|
|
Example : |
241
|
|
|
|
|
|
|
Returns : string |
242
|
|
|
|
|
|
|
Args : |
243
|
|
|
|
|
|
|
|
244
|
|
|
|
|
|
|
|
245
|
|
|
|
|
|
|
=cut |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
sub version { |
248
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
249
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
250
|
|
|
|
|
|
|
} |
251
|
|
|
|
|
|
|
|
252
|
|
|
|
|
|
|
=head2 revision |
253
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
Title : revision |
255
|
|
|
|
|
|
|
Usage : my $revision= $header->revision; |
256
|
|
|
|
|
|
|
Function: Returns the revision of the file being parsed if such exists |
257
|
|
|
|
|
|
|
Throws : |
258
|
|
|
|
|
|
|
Example : |
259
|
|
|
|
|
|
|
Returns : string |
260
|
|
|
|
|
|
|
Args : |
261
|
|
|
|
|
|
|
|
262
|
|
|
|
|
|
|
|
263
|
|
|
|
|
|
|
=cut |
264
|
|
|
|
|
|
|
|
265
|
|
|
|
|
|
|
sub revision { |
266
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
267
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
268
|
|
|
|
|
|
|
} |
269
|
|
|
|
|
|
|
|
270
|
|
|
|
|
|
|
=head2 _check |
271
|
|
|
|
|
|
|
|
272
|
|
|
|
|
|
|
Title : _check |
273
|
|
|
|
|
|
|
Usage : if ($self->_check('weights') { #do something} else {return 0;} |
274
|
|
|
|
|
|
|
Function: Checks if the method called is aplicable to the file format |
275
|
|
|
|
|
|
|
Throws : |
276
|
|
|
|
|
|
|
Example : |
277
|
|
|
|
|
|
|
Returns : boolean |
278
|
|
|
|
|
|
|
Args : string |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
|
281
|
|
|
|
|
|
|
=cut |
282
|
|
|
|
|
|
|
|
283
|
|
|
|
|
|
|
sub _check { |
284
|
0
|
|
|
0
|
|
|
my $self = shift; |
285
|
0
|
|
|
|
|
|
$self->throw_not_implemented(); |
286
|
|
|
|
|
|
|
} |
287
|
|
|
|
|
|
|
|
288
|
|
|
|
|
|
|
1; |