annotate dir_plugins/miRNA.pm @ 10:f594c6bed58f draft default tip

Uploaded
author dvanzessen
date Tue, 21 Apr 2020 11:40:19 +0000
parents e545d0a25ffe
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
1 =head1 LICENSE
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
2
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
3 Copyright [1999-2015] Wellcome Trust Sanger Institute and the EMBL-European Bioinformatics Institute
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
4 Copyright [2016-2018] EMBL-European Bioinformatics Institute
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
5
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
6 Licensed under the Apache License, Version 2.0 (the "License");
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
7 you may not use this file except in compliance with the License.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
8 You may obtain a copy of the License at
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
9
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
10 http://www.apache.org/licenses/LICENSE-2.0
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
11
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
12 Unless required by applicable law or agreed to in writing, software
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
13 distributed under the License is distributed on an "AS IS" BASIS,
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
14 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
15 See the License for the specific language governing permissions and
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
16 limitations under the License.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
17
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
18 =head1 CONTACT
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
19
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
20 Ensembl <http://www.ensembl.org/info/about/contact/index.html>
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
21
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
22 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
23
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
24 =head1 NAME
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
25
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
26 miRNA
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
27
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
28 =head1 SYNOPSIS
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
29
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
30 mv miRNA.pm ~/.vep/Plugins
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
31 ./vep -i variations.vcf --plugin miRNA
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
32
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
33 =head1 DESCRIPTION
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
34
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
35 A VEP plugin that determines where in the secondary structure of a miRNA a
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
36 variant falls.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
37
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
38 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
39
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
40 package miRNA;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
41
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
42 use strict;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
43 use warnings;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
44
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
45 use Bio::EnsEMBL::Variation::Utils::BaseVepPlugin;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
46
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
47 use base qw(Bio::EnsEMBL::Variation::Utils::BaseVepPlugin);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
48
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
49 sub feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
50 return ['Transcript'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
51 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
52
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
53 sub get_header_info {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
54 my $self = shift;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
55 return {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
56 miRNA => 'SO term for miRNA component containing the variant'
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
57 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
58 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
59
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
60 sub run {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
61 my ($self, $tva) = @_;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
62
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
63 my $tv = $tva->transcript_variation;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
64 my $tr = $tva->transcript;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
65
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
66 # obviously this only works for *RNA transcripts
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
67 return {} unless $tr->biotype =~ /RNA/;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
68
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
69 # and it only works if the TV falls in the cDNA
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
70 return {} unless $tv->cdna_start && $tv->cdna_end;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
71
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
72 # get attribute if already cached
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
73 my ($attrib) = @{$tr->get_all_Attributes('ncRNA')};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
74
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
75 # bit of a cheat to get attrib if ncRNA attribute hasn't been cached
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
76 if(!$attrib && defined($self->{config}->{ta})) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
77 delete $tr->{attributes};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
78 $tr->{adaptor} = $self->{config}->{ta};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
79 ($attrib) = @{$tr->get_all_Attributes('ncRNA')};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
80 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
81
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
82 return {} unless $attrib;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
83
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
84 # split out string to get coords and structure string
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
85 my ($start, $end, $struct) = split /\s+|\:/, $attrib->value;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
86 return {} unless $struct && $struct =~ /[\(\.\)]+/;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
87
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
88 # variant not in given structure?
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
89 return { miRNA => 'None' } unless $tv->cdna_start <= $end && $tv->cdna_end >= $start;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
90
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
91 # parse out structure
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
92 my @struct;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
93 while($struct =~ m/([\.\(\)])([0-9]+)?/g) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
94 my $num = $2 || 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
95 push @struct, $1 for(1..$num);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
96 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
97
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
98 # get struct element types overlapped by variant
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
99 my %chars;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
100 for my $pos($tv->cdna_start..$tv->cdna_end) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
101 $pos -= $start;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
102 next if $pos < 0 or $pos > scalar @struct;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
103 $chars{$struct[$pos]} = 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
104 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
105
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
106 # map element types to SO terms
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
107 my %map = (
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
108 '(' => 'miRNA_stem',
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
109 ')' => 'miRNA_stem',
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
110 '.' => 'miRNA_loop'
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
111 );
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
112
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
113 return {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
114 miRNA => join(",", sort map {$map{$_}} keys %chars)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
115 };
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
116 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
117
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
118 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
119