annotate dir_plugins/LOVD.pm @ 3:49397129aec0 draft

Uploaded
author dvanzessen
date Mon, 15 Jul 2019 05:20:39 -0400
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 LOVD
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 LOVD.pm ~/.vep/Plugins
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
31 ./vep -i variations.vcf --plugin LOVD
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 retrieves LOVD variation data from http://www.lovd.nl/.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
36
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
37 Please be aware that LOVD is a public resource of curated variants, therefore
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
38 please respect this resource and avoid intensive querying of their databases
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
39 using this plugin, as it will impact the availability of this resource for
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
40 others.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
41
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
42 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
43
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
44 package LOVD;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
45
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
46 use strict;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
47 use warnings;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
48 use LWP::UserAgent;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
49
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
50 use Bio::EnsEMBL::Variation::Utils::BaseVepPlugin;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
51
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
52 use base qw(Bio::EnsEMBL::Variation::Utils::BaseVepPlugin);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
53
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
54 sub version {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
55 return '2.5';
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
56 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
57
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
58 sub feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
59 return ['Transcript'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
60 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
61
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
62 sub get_header_info {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
63 return {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
64 LOVD => "LOVD variant ID",
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
65 };
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
66 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
67
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
68 sub run {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
69 my ($self, $tva) = @_;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
70
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
71 $self->{has_cache} = 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
72
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
73 my %hg_assembly = ( 'grch37' => 'hg19', 'grch38' => 'hg38' );
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
74 my $assembly = lc($self->{config}->{assembly});
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
75
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
76 # only works on human
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
77 die("ERROR: LOVD plugin works only on human data") unless $self->{config}->{species} =~ /human|homo/i;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
78
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
79 # only work on 2 human assemblies
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
80 die("ERROR: LOVD plugin works only on human assemblies ".join(' and ', keys(%hg_assembly))) unless $hg_assembly{$assembly};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
81
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
82 # get the VF object
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
83 my $vf = $tva->variation_feature;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
84 return {} unless defined $vf;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
85
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
86 # set up a LWP UserAgent
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
87 my $ua = LWP::UserAgent->new;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
88 $ua->env_proxy;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
89
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
90 my $chr = $vf->{chr};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
91 $chr =~ s/^chr//;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
92
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
93 my $locus = sprintf('chr%s:%s_%s', $chr, $vf->{start}, $vf->{end});
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
94
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
95 my $data;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
96
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
97 # check the cache
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
98 if(!exists($self->{lovd_cache}->{$locus})) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
99
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
100 # construct a LOVD URL
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
101 my $url = sprintf('http://www.lovd.nl/search.php?build=%s&position=%s', $hg_assembly{$assembly}, $locus);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
102
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
103 # get the accession (only need the head to get the redirect URL that contains the accession)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
104 my $response = $ua->get($url);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
105
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
106 if($response->is_success) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
107
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
108 # parse the data into a hash
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
109 for(grep {$_ !~ /hg_build/} split /\cJ/, $response->decoded_content) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
110 s/\"//g;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
111
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
112 my ($build, $pos, $gene, $acc, $dna, $url) = split /\t/;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
113
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
114 $data = join(',', $gene, $acc, $dna)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
115 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
116
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
117 $self->{lovd_cache}->{$locus} = $data;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
118 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
119 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
120 else {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
121 $data = $self->{lovd_cache}->{$locus};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
122 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
123
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
124 return $data ? { LOVD => $data } : {};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
125 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
126
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
127 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
128