annotate dir_plugins/CSN.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 CSN
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 CSN.pm ~/.vep/Plugins
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
31 ./vep -i variations.vcf --cache --plugin CSN
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 This is a plugin for the Ensembl Variant Effect Predictor (VEP) that
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
36 reports Clinical Sequencing Nomenclature (CSN) for variants.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
37
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
38 Each notation is given with reference to the transcript identifier;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
39 specify "--plugin CSN,1" to remove this identifier from the CSN string.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
40
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
41 You may also wish to specify "--no_escape" to prevent the "=" in "p.="
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
42 notations being converted to the URI-escaped equivalent "p.%3D"; doing
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
43 so may break parsers looking for "=" as a KEY=VALUE separator.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
44
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
45 See http://biorxiv.org/content/early/2015/03/21/016808.1
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
46
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
47 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
48
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
49 package CSN;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
50
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
51 use strict;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
52 use warnings;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
53
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
54 use Bio::EnsEMBL::Variation::Utils::BaseVepPlugin;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
55 use Bio::EnsEMBL::Variation::DBSQL::TranscriptVariationAdaptor;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
56 use Bio::EnsEMBL::Variation::DBSQL::DBAdaptor;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
57
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
58 use base qw(Bio::EnsEMBL::Variation::Utils::BaseVepPlugin);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
59
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
60 sub new {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
61 my $class = shift;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
62
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
63 my $self = $class->SUPER::new(@_);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
64
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
65 # check config is OK
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
66
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
67 # FASTA file defined, optimal
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
68 if(!defined($self->{config}->{fasta})) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
69
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
70 # offline mode won't work without FASTA
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
71 die("ERROR: Cannot generate CSN without either a FASTA file (--fasta) or a database connection (--cache or --database)\n") if defined($self->{config}->{offline}) and !defined($self->{config}->{quiet});
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
72
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
73 # cache mode will work, but DB will be accessed
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
74 warn("WARNING: Database will be accessed using this plugin; use a FASTA file (--fasta) for optimal performance") if defined($self->{config}->{cache}) and !defined($self->{config}->{quiet});
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
75 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
76
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
77 no warnings 'once';
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
78 $Bio::EnsEMBL::Variation::DBSQL::TranscriptVariationAdaptor::DEFAULT_SHIFT_HGVS_VARIANTS_3PRIME = 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
79 no warnings 'once';
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
80 $Bio::EnsEMBL::Variation::DBSQL::DBAdaptor::DEFAULT_SHIFT_HGVS_VARIANTS_3PRIME = 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
81
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
82 $self->{remove_transcript_ID} = $self->params->[0];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
83
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
84 return $self;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
85 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
86
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
87 sub feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
88 return ['Transcript'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
89 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
90
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
91 sub variant_feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
92 return ['VariationFeature'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
93 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
94
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
95 sub get_header_info {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
96 return { CSN => 'Clinical Sequencing Nomenclature'};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
97 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
98
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
99 sub run {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
100 my ($self, $tva) = @_;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
101
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
102 my ($hgvs_c, $hgvs_p) = ($tva->hgvs_transcript || '', $tva->hgvs_protein || '');
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
103
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
104 return {} unless $hgvs_c;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
105
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
106 # trim off transcript/protein ID
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
107 $hgvs_c =~ s/.+\:// if $self->{remove_transcript_ID};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
108 $hgvs_p =~ s/.+\://;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
109
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
110 # change Ter to X
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
111 $hgvs_p =~ s/Ter/X/g;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
112
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
113 # leave just p.=
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
114 $hgvs_p = 'p.=' if $hgvs_p =~ /p\.\=/;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
115
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
116 # escape
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
117 $hgvs_p =~ s/\=/\%3D/g unless $self->{config}->{no_escape};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
118
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
119 return { CSN => $hgvs_c.($hgvs_p ? '_'.$hgvs_p : '') };
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
120 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
121
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
122 1;