annotate variant_effect_predictor/Bio/Tools/RepeatMasker.pm @ 0:21066c0abaf5 draft

Uploaded
author willmclaren
date Fri, 03 Aug 2012 10:04:48 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
1 # BioPerl module for Bio::Tools::RepeatMasker
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
2 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
3 # Cared for by Shawn Hoon <shawnh@fugu-sg.org>
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
4 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
5 # Copyright Shawn Hoon
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
6 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
7 # You may distribute this module under the same terms as perl itself
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
8
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
9 # POD documentation - main docs before the code
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
10
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
11 =head1 NAME
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
12
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
13 Bio::Tools::RepeatMasker - DESCRIPTION of Object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
14
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
15 =head1 SYNOPSIS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
16
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
17 use Bio::Tools::RepeatMasker;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
18 my $parser = new Bio::Tools::RepeatMasker(-file => 'seq.fa.out');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
19 while( my $result = $parser->next_result ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
20
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
21 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
22
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
23 =head1 DESCRIPTION
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
24
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
25 A parser for RepeatMasker output
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
26
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
27 =head1 FEEDBACK
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
28
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
29 =head2 Mailing Lists
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
30
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
31 User feedback is an integral part of the evolution of this and other
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
32 Bioperl modules. Send your comments and suggestions preferably to
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
33 the Bioperl mailing list. Your participation is much appreciated.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
34
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
35 bioperl-l@bioperl.org - General discussion
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
36 http://bioperl.org/MailList.shtml - About the mailing lists
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
37
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
38 =head2 Reporting Bugs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
39
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
40 Report bugs to the Bioperl bug tracking system to help us keep track
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
41 of the bugs and their resolution. Bug reports can be submitted via
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
42 email or the web:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
43
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
44 bioperl-bugs@bioperl.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
45 http://bugzilla.bioperl.org/
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
46
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
47 =head1 AUTHOR - Shawn Hoon
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
48
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
49 Email shawnh@fugu-sg.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
50
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
51 Describe contact details here
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
52
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
53 =head1 CONTRIBUTORS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
54
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
55 Additional contributors names and emails here
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
56
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
57 =head1 APPENDIX
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
58
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
59 The rest of the documentation details each of the object methods.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
60 Internal methods are usually preceded with a _
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
61
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
62 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
63
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
64
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
65 # Let the code begin...
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
66
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
67
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
68 package Bio::Tools::RepeatMasker;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
69 use vars qw(@ISA);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
70 use strict;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
71
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
72 use Bio::Root::Root;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
73 use Bio::SeqFeature::FeaturePair;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
74 use Bio::Root::IO;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
75
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
76 @ISA = qw(Bio::Root::Root Bio::Root::IO );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
77
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
78 =head2 new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
79
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
80 Title : new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
81 Usage : my $obj = new Bio::Tools::RepeatMasker();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
82 Function: Builds a new Bio::Tools::RepeatMasker object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
83 Returns : Bio::Tools::RepeatMasker
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
84 Args : -fh/-file => $val, # for initing input, see Bio::Root::IO
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
85
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
86
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
87 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
88
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
89 sub new {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
90 my($class,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
91
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
92 my $self = $class->SUPER::new(@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
93 $self->_initialize_io(@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
94
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
95 return $self;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
96 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
97
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
98 =head2 next_result
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
99
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
100 Title : next_result
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
101 Usage : my $r = $rpt_masker->next_result
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
102 Function: Get the next result set from parser data
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
103 Returns : L<Bio::SeqFeature::FeaturePair>
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
104 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
105
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
106
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
107 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
108
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
109 sub next_result{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
110 my ($self) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
111 while ($_=$self->_readline()) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
112 if (/no repetitive sequences detected/) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
113 print STDERR "RepeatMasker didn't find any repetitive sequences\n";
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
114 return ;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
115 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
116 if (/\d+/) { #ignore introductory lines
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
117 my @element = split;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
118 # ignore features with negatives
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
119 next if ($element[11-13] =~ /-/);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
120 my (%feat1, %feat2);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
121 my ($score, $query_name, $query_start, $query_end, $strand,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
122 $repeat_name, $repeat_class ) = (split)[0, 4, 5, 6, 8, 9, 10];
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
123
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
124 my ($hit_start,$hit_end);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
125 if ($strand eq '+') {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
126 ($hit_start, $hit_end) = (split)[11, 12];
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
127 $strand = 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
128 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
129 elsif ($strand eq 'C') {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
130 ($hit_start, $hit_end) = (split)[12, 13];
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
131 $strand = -1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
132 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
133 my $rf = Bio::SeqFeature::Generic->new;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
134 $rf->seq_id ($query_name);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
135 $rf->score ($score);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
136 $rf->start ($query_start);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
137 $rf->end ($query_end);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
138 $rf->strand ($strand);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
139 $rf->source_tag ("RepeatMasker");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
140 $rf->primary_tag ($repeat_class);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
141 my $rf2 = Bio::SeqFeature::Generic->new;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
142 $rf2->seq_id ($repeat_name);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
143 $rf2->score ($score);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
144 $rf2->start ($hit_start);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
145 $rf2->end ($hit_end);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
146 $rf2->strand ($strand);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
147 $rf2->source_tag ("RepeatMasker");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
148 $rf->primary_tag ($repeat_class);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
149 my $fp = Bio::SeqFeature::FeaturePair->new(-feature1=>$rf,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
150 -feature2=>$rf2);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
151
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
152 return $fp;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
153 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
154 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
155 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
156
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
157 1;