annotate variant_effect_predictor/Bio/EnsEMBL/PaddedSlice.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 =head1 LICENSE
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
2
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
3 Copyright (c) 1999-2012 The European Bioinformatics Institute and
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4 Genome Research Limited. All rights reserved.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
6 This software is distributed under a modified Apache license.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
7 For license details, please see
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
8
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
9 http://www.ensembl.org/info/about/code_licence.html
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
10
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
11 =head1 CONTACT
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
12
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
13 Please email comments or questions to the public Ensembl
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
14 developers list at <dev@ensembl.org>.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
15
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
16 Questions may also be sent to the Ensembl help desk at
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
17 <helpdesk@ensembl.org>.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
18
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
19 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
20
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
21 =head1 NAME
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
22
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
23 Bio::EnsEMBL::PaddedSlice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
24
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
25 =head1 DESCRIPTION
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
26
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
27 Used when dumping Slices which represet a portion of the sequence region
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
28 they map to e.g. the first section of human Y. The code will return N
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
29 as sequence if an attempt is made to retrieve sequence not covered by the
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
30 Slice given. This makes the code very memory efficient if sequence dumping
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
31 is carried out using C<subseq()> calls.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
32
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
33 =head1 METHODS
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
34
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
35 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
36
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
37 package Bio::EnsEMBL::PaddedSlice;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
38
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
39 use Bio::EnsEMBL::Utils::Argument qw/rearrange/;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
40 use Bio::EnsEMBL::Utils::Scalar qw/assert_ref assert_strand/;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
41 use base qw/Bio::EnsEMBL::Utils::Proxy/;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
42
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
43 =head2 new()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
44
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
45 Arg [SLICE] : The Slice to proxy
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
46 Example : my $newobj = Bio::EnsEMBL::PaddedSlice->new($myobj);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
47 Description : Provides a new instance of a padded slice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
48 Returntype : Bio::EnsEMBL::PaddedSlice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
49 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
50 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
51 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
52
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
53 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
54
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
55 sub new {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
56 my ($class, @args) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
57 my ($slice) = rearrange([qw/slice/], @args);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
58 return $class->SUPER::new($slice);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
59 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
60
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
61 =head2 start()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
62
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
63 Example : $slice->start();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
64 Description : Always returns 1 since all padded slices start at 1
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65 Returntype : Int
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72 sub start {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73 my ($self) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74 return 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 =head2 end()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 Example : $slice->end();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80 Description : Always returns the backing slice sequence region length
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81 Returntype : Int
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
86 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
87
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
88 sub end {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89 my ($self) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90 return $self->seq_region_length();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
93 =head2 length()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
94
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
95 Example : $slice->length();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
96 Description : Delegates to C<end()>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 Returntype : Int
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
104 sub length {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
105 my ($self) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
106 return $self->end();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
107 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109 =head2 seq()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 Example : my $seq = $slice->seq()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112 Description : Returns the entire sequence of the backing slice but padded
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 with N's at the beginning and the end of the slice where
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114 applicable
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115 Returntype : Scalar string
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122 sub seq {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123 my ($self) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124 my $parent_slice = $self->__proxy();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125 my $pad_start = 'N' x ( $parent_slice->start() - 1 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126 my $pad_end = 'N' x ( $parent_slice->seq_region_length() - $parent_slice->end() );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127 my $seq = $parent_slice->seq();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128 return $pad_start . $seq . $pad_end;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131 =head2 subseq()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133 Arg [1] : Int; start position of the subslice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
134 Arg [2] : Int; end position of the subslice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
135 Arg [3] : Int; strand of the subslice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
136 Example : my $subseq = $slice->subseq(1, 1_000_000);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
137 Description : Returns a portion of the sequence padded with N's if required
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 Returntype : Scalar string
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141 Status : -
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 sub subseq {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146 my ( $self, $start, $end, $strand ) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148 if ( $end+1 < $start ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149 throw("End coord + 1 is less than start coord");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 return '' if( $start == $end + 1);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154 $strand = 1 unless(defined $strand);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
155 assert_strand($strand, 'strand');
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
156
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
157 my $parent_slice = $self->__proxy();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
158
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
159 #Coords relative to the SeqRegion i.e. huge
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
160 my $parent_start = $parent_slice->start();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
161 my $parent_end = $parent_slice->end();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
162
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
163 #Return if we were upstream of overlap
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
164 if($start < $parent_start && $end < $parent_start) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
165 return N x (( $end - $start )+1);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
166 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
167 #Return if we were downstream of overlap
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
168 if($start > $parent_end && $end > $parent_end) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
169 return N x (( $end - $start )+1);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
170 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
171
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
172 my $prefix = '';
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
173 my $suffix = '';
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
174 my $subslice_start = ($start - $parent_start)+1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
175 my $subslice_end = ($end - $parent_start) + 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
176 if($start < $parent_start) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
177 $prefix = N x ($parent_start - $start);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
178 $subslice_start = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
179 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
180 if($end > $parent_end) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
181 $suffix = N x ($end - $parent_end);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
182 $subslice_end = (($parent_end - $parent_start)+1);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
183 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
184
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
185 my $subseq = $parent_slice->subseq($subslice_start, $subslice_end, $strand);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
186
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
187 return $prefix . $subseq . $suffix;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
188 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
189
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
190 =head2 subseq()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
191
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
192 Arg [1] : Int; start position of the subslice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
193 Arg [2] : Int; end position of the subslice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
194 Arg [3] : Int; strand of the subslice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
195 Example : my $subseq = $slice->subseq(1, 1_000_000);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
196 Description : Returns a portion of the sequence padded with N's if required
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
197 Returntype : Scalar string
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
198 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
199 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
200 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
201
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
202 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
203
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
204 sub sub_Slice {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
205 die "Unsupported";
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
206 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
207
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
208
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
209 =head2 __resolver()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
210
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
211 Description : Delegates all non-overriden actions onto the backing slice
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
212 Returntype : CodeRef
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
213 Exceptions : None
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
214 Caller : public
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
215 Status : -
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
216
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
217 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
218
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
219 sub __resolver {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
220 my ($self, $package_name, $method) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
221 return sub {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
222 my ($local_self, @args) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
223 return $local_self->__proxy()->$method(@args);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
224 };
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
225 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
226
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
227
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
228 1;