annotate variant_effect_predictor/Bio/Tools/Genomewise.pm @ 0:2bc9b66ada89 draft default tip

Uploaded
author mahtabm
date Thu, 11 Apr 2013 06:29:17 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
1 # $Id: Genomewise.pm,v 1.1.2.1 2003/03/25 12:32:16 heikki Exp $
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
2 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
3 # BioPerl module for Bio::Tools::Genomewise
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5 # Copyright Jason Stajich <jason@bioperl.org>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
6 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
7 # You may distribute this module under the same terms as perl itself
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
8 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
9 # POD documentation - main docs before the code
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
10
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
11 =head1 NAME
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
12
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
13 Bio::Tools::Genomewise - Results of one Genomewise run
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
14
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
15 =head1 SYNOPSIS
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
16
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
17 use Bio::Tools::Genomewise;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
18 my $gw = Bio::Tools::Genomewise(-file=>"genomewise.out");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
19
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
20 while (my $gene = $gw->next_prediction){
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
21 my @transcripts = $gw->transcripts;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
22 foreach my $t(@transcripts){
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
23 my @exons = $t->exons;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
24 foreach my $e(@exons){
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
25 print $e->start." ".$e->end."\n";
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
26 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
27 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
28 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
29
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
30 =head1 DESCRIPTION
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
31
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
32 This is the parser for the output of Genewise. It takes either a file
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
33 handle or a file name and returns a
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
34 Bio::SeqFeature::Gene::GeneStructure object. You will need to specify
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
35 the proper target sequence id on the object with the
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
36 $feature-E<gt>seq_id($seqid).
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
37
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
38 =head1 FEEDBACK
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
39
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
40 =head2 Mailing Lists
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
41
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
42 User feedback is an integral part of the evolution of this and other
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
43 Bioperl modules. Send your comments and suggestions preferably to one
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
44 of the Bioperl mailing lists. Your participation is much appreciated.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
45
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
46 bioperl-l@bioperl.org - General discussion
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
47 http://bio.perl.org/MailList.html - About the mailing lists
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
48
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
49 =head2 Reporting Bugs
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
50
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
51 Report bugs to the Bioperl bug tracking system to help us keep track
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
52 the bugs and their resolution. Bug reports can be submitted via email
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
53 or the web:
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
54
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
55 bioperl-bugs@bio.perl.org
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
56 http://bugzilla.bioperl.org/
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
57
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
58 =head1 AUTHOR - Fugu Team
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
59
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
60 Email: fugui@worf.fugu-sg.org
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
61
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
62 =head1 APPENDIX
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
63
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
64 The rest of the documentation details each of the object methods. Internal methods are usually preceded with a _
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69 # Let the code begin...
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72 package Bio::Tools::Genomewise;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73 use vars qw(@ISA $Srctag);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74 use strict;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76 use Bio::Tools::Genewise;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 use Bio::Tools::AnalysisResult;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78 use Bio::SeqFeature::Generic;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 use Bio::SeqFeature::Gene::Exon;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80 use Bio::SeqFeature::FeaturePair;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81 use Bio::SeqFeature::Gene::Transcript;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82 use Bio::SeqFeature::Gene::GeneStructure;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84 @ISA = qw(Bio::Tools::Genewise);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
86 $Srctag = 'genomewise';
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
87
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
88 =head2 new
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90 Title : new
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 Usage : $obj->new(-file=>"genewise.out");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92 $obj->new(-fh=>\*GW);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
93 Function: Constructor for genomewise wrapper. Takes either a file or filehandle
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
94 Example :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
95 Returns : L<Bio::Tools::Genomewise>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
96
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99 sub new {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100 my($class,@args) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101 my $self = $class->SUPER::new(@args);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102 return $self;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
104
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
105 =head2 _get_strand
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
106
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
107 Title : _get_strand
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108 Usage : $obj->_get_strand
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109 Function: takes start and end values, swap them if start>end and returns end
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110 Example :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 Returns :$start,$end,$strand
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115 =head2 score
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 Title : score
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118 Usage : $obj->score
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119 Function: get/set for score info
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120 Example :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121 Returns : a score value
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125 =head2 _prot_id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127 Title : _prot_id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128 Usage : $obj->_prot_id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129 Function: get/set for protein id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130 Example :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131 Returns :a protein id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
134
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
135 =head2 _target_id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
136
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
137 Title : _target_id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 Usage : $obj->_target_id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139 Function: get/set for genomic sequence id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140 Example :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141 Returns :a target id
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
142
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
143 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
144
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
145
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146 =head2 next_prediction
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148 Title : next_prediction
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149 Usage : while($gene = $genewise->next_prediction()) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 # do something
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 Function: Returns the gene structure prediction of the Genomewise result
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153 file. Call this method repeatedly until FALSE is returned.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
155 Example :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
156 Returns : a Bio::SeqFeature::Gene::GeneStructure object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
157 Args :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
158
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
159 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
160
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
161
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
162 sub next_prediction {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
163 my ($self) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
164
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
165 my $genes;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
166 while ($_ = $self->_readline) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
167 $self->debug( $_ ) if( $self->verbose > 0);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
168 last if( /^\/\//);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
169
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
170 if( /^Gene\s+\d+\s*$/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
171 $genes = new Bio::SeqFeature::Gene::GeneStructure
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
172 (-source => $Srctag,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
173 -seq_id => $self->_target_id, # if this had been specified
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
174 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
175 $_ = $self->_readline;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
176 $self->debug( $_ ) if( $self->verbose > 0);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
177
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
178 unless ( /^Gene\s+(\d+)\s+(\d+)\s*$/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
179 $self->warn("Unparseable genomewise output");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
180 last;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
181 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
182 my $transcript = new Bio::SeqFeature::Gene::Transcript
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
183 (-source => $Srctag,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
184 -seq_id => $self->_target_id, # if this had been specified
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
185 -start => $1,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
186 -end => $2,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
187 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
188 my $nbr = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
189 while( $_ = $self->_readline ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
190 $self->debug( $_ ) if( $self->verbose > 0);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
191
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
192 unless( m/^\s+Exon\s+(\d+)\s+(\d+)\s+phase\s+(\d+)/ ){
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
193 $self->_pushback($_);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
194 last;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
195 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
196 my ($e_start,$e_end,$phase,$e_strand) = ($1,$2,$3);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
197
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
198 ($e_start,$e_end,$e_strand) = $self->_get_strand($e_start,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
199 $e_end);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
200 $transcript->strand($e_strand) unless $transcript->strand != 0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
201
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
202 my $exon = new Bio::SeqFeature::Gene::Exon
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
203 (-seq_id=>$self->_target_id,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
204 -source => $Srctag,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
205 -start=>$e_start,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
206 -end=>$e_end,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
207 -frame => $phase,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
208 -strand=>$e_strand);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
209 $exon->add_tag_value("Exon",$nbr++);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
210 $exon->add_tag_value('phase',$phase);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
211 $transcript->add_exon($exon);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
212 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
213 $genes->add_transcript($transcript);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
214 last; # only process a single gene at a time
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
215 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
216 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
217 return $genes;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
218 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
219 1;