annotate dir_plugins/SameCodon.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 SameCodon
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 SameCodon.pm ~/.vep/Plugins
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
31 ./vep -i variations.vcf --plugin SameCodon
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 reports existing variants that fall in the same codon.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
36
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
37 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
38
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
39 package SameCodon;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
40
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
41 use strict;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
42 use warnings;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
43
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
44 use Bio::EnsEMBL::Variation::Utils::BaseVepPlugin;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
45
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
46 use base qw(Bio::EnsEMBL::Variation::Utils::BaseVepPlugin);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
47
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
48 sub version {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
49 return '3.0';
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
50 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
51
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
52 sub feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
53 return ['Transcript'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
54 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
55
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
56 sub get_header_info {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
57 return {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
58 SameCodon => "Existing variant IDs that fall in the same codon",
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
59 };
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
60 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
61
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
62 sub run {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
63
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
64 my ($self, $tva) = @_;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
65
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
66 if ($self->config->{offline}) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
67 die "A connection to the database is required to use the plugin SameCodon\n";
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
68 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
69
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
70 my $tv = $tva->transcript_variation;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
71 my $vf = $tv->variation_feature;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
72 my ($pep_start, $pep_end) = ($tv->translation_start, $tv->translation_end);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
73 my ($vf_start, $vf_end) = ($vf->start, $vf->end);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
74
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
75 return {} unless defined($pep_start) && defined($pep_end);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
76
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
77 my $config = $self->{config};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
78
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
79 # we need to map the TV start and end coords to the genome
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
80 # needs to be done through the mapper in case the codon spans exons
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
81 my $mapper = $tv->_mapper();
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
82
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
83 return {} unless defined($mapper);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
84
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
85 my @coords = $mapper->pep2genomic($pep_start, $pep_end);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
86
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
87 return {} unless scalar @coords;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
88 return {} if grep {!$_->isa('Bio::EnsEMBL::Mapper::Coordinate')} @coords;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
89
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
90 my @results;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
91 # we might get multiple "slices" if the codon that the variant falls in spans exons
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
92 foreach my $coord(@coords) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
93
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
94 my ($slice_start, $slice_end) = ($coord->start, $coord->end);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
95
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
96 my $sub_slice = $vf->slice->sub_Slice($slice_start, $slice_end);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
97 my $vf_adaptor = $vf->slice->_get_VariationFeatureAdaptor();
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
98 push @results,
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
99 map {$_->variation_name}
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
100 grep {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
101 $_->variation_name ne $vf->variation_name &&
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
102 $_->seq_region_start != $vf_start &&
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
103 $_->seq_region_end != $vf_end &&
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
104 scalar $mapper->genomic2cds($_->seq_region_start, $_->seq_region_end, 1) >= 1
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
105 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
106 @{$vf_adaptor->fetch_all_by_Slice_SO_terms($sub_slice)};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
107 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
108
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
109 return {} unless scalar @results;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
110
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
111 return {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
112 SameCodon => join ",", @results
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
113 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
114 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
115
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
116 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
117