annotate variant_effect_predictor/Bio/Tools/RepeatMasker.pm @ 3:d30fa12e4cc5 default tip

Merge heads 2:a5976b2dce6f and 1:09613ce8151e which were created as a result of a recently fixed bug.
author devteam <devteam@galaxyproject.org>
date Mon, 13 Jan 2014 10:38:30 -0500
parents 1f6dce3d34e0
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
1 # BioPerl module for Bio::Tools::RepeatMasker
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
2 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
3 # Cared for by Shawn Hoon <shawnh@fugu-sg.org>
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
4 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
5 # Copyright Shawn Hoon
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
6 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
7 # You may distribute this module under the same terms as perl itself
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
8
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
9 # POD documentation - main docs before the code
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
10
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
11 =head1 NAME
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
12
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
13 Bio::Tools::RepeatMasker - DESCRIPTION of Object
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
14
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
15 =head1 SYNOPSIS
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
16
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
17 use Bio::Tools::RepeatMasker;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
18 my $parser = new Bio::Tools::RepeatMasker(-file => 'seq.fa.out');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
19 while( my $result = $parser->next_result ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
20
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
21 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
22
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
23 =head1 DESCRIPTION
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
24
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
25 A parser for RepeatMasker output
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
26
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
27 =head1 FEEDBACK
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
28
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
29 =head2 Mailing Lists
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
30
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
31 User feedback is an integral part of the evolution of this and other
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
32 Bioperl modules. Send your comments and suggestions preferably to
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
33 the Bioperl mailing list. Your participation is much appreciated.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
34
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
35 bioperl-l@bioperl.org - General discussion
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
36 http://bioperl.org/MailList.shtml - About the mailing lists
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
37
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
38 =head2 Reporting Bugs
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
39
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
40 Report bugs to the Bioperl bug tracking system to help us keep track
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
41 of the bugs and their resolution. Bug reports can be submitted via
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
42 email or the web:
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
43
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
44 bioperl-bugs@bioperl.org
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
45 http://bugzilla.bioperl.org/
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
46
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
47 =head1 AUTHOR - Shawn Hoon
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
48
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
49 Email shawnh@fugu-sg.org
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
50
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
51 Describe contact details here
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
52
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
53 =head1 CONTRIBUTORS
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
54
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
55 Additional contributors names and emails here
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
56
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
57 =head1 APPENDIX
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
58
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
59 The rest of the documentation details each of the object methods.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
60 Internal methods are usually preceded with a _
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
61
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
62 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
63
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
64
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
65 # Let the code begin...
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
66
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
67
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
68 package Bio::Tools::RepeatMasker;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
69 use vars qw(@ISA);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
70 use strict;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
71
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
72 use Bio::Root::Root;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
73 use Bio::SeqFeature::FeaturePair;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
74 use Bio::Root::IO;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
75
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
76 @ISA = qw(Bio::Root::Root Bio::Root::IO );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
77
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
78 =head2 new
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
79
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
80 Title : new
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
81 Usage : my $obj = new Bio::Tools::RepeatMasker();
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
82 Function: Builds a new Bio::Tools::RepeatMasker object
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
83 Returns : Bio::Tools::RepeatMasker
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
84 Args : -fh/-file => $val, # for initing input, see Bio::Root::IO
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
85
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
86
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
87 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
88
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
89 sub new {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
90 my($class,@args) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
91
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
92 my $self = $class->SUPER::new(@args);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
93 $self->_initialize_io(@args);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
94
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
95 return $self;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
96 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
97
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
98 =head2 next_result
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
99
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
100 Title : next_result
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
101 Usage : my $r = $rpt_masker->next_result
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
102 Function: Get the next result set from parser data
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
103 Returns : L<Bio::SeqFeature::FeaturePair>
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
104 Args : none
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
105
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
106
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
107 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
108
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
109 sub next_result{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
110 my ($self) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
111 while ($_=$self->_readline()) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
112 if (/no repetitive sequences detected/) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
113 print STDERR "RepeatMasker didn't find any repetitive sequences\n";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
114 return ;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
115 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
116 if (/\d+/) { #ignore introductory lines
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
117 my @element = split;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
118 # ignore features with negatives
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
119 next if ($element[11-13] =~ /-/);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
120 my (%feat1, %feat2);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
121 my ($score, $query_name, $query_start, $query_end, $strand,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
122 $repeat_name, $repeat_class ) = (split)[0, 4, 5, 6, 8, 9, 10];
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
123
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
124 my ($hit_start,$hit_end);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
125 if ($strand eq '+') {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
126 ($hit_start, $hit_end) = (split)[11, 12];
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
127 $strand = 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
128 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
129 elsif ($strand eq 'C') {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
130 ($hit_start, $hit_end) = (split)[12, 13];
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
131 $strand = -1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
132 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
133 my $rf = Bio::SeqFeature::Generic->new;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
134 $rf->seq_id ($query_name);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
135 $rf->score ($score);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
136 $rf->start ($query_start);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
137 $rf->end ($query_end);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
138 $rf->strand ($strand);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
139 $rf->source_tag ("RepeatMasker");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
140 $rf->primary_tag ($repeat_class);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
141 my $rf2 = Bio::SeqFeature::Generic->new;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
142 $rf2->seq_id ($repeat_name);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
143 $rf2->score ($score);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
144 $rf2->start ($hit_start);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
145 $rf2->end ($hit_end);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
146 $rf2->strand ($strand);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
147 $rf2->source_tag ("RepeatMasker");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
148 $rf->primary_tag ($repeat_class);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
149 my $fp = Bio::SeqFeature::FeaturePair->new(-feature1=>$rf,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
150 -feature2=>$rf2);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
151
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
152 return $fp;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
153 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
154 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
155 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
156
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
157 1;